This is what I always believed for years- that to unmask one needed a fast recovery speed and minimal or no iron rejection (any form of discrimination). Since then I've seen with my own eyes and ears in numerous head to heads in the field that there are more than one way to skin that unmasking cat. Comparing a machine with a faster recovery speed and low or no iron rejection being set, to that of one with built in iron rejection and a slow recovery speed...Both machines have seen badly masked coins before digging just as well as each other. And I'm talking some coins so badly masked that they were a complete null all the way around them (even for the low iron rejection setup) and would only sing "COIN" from one very tight angle for both machines.
The difference? The slow machine is using a quality DD coil with a super sharp DD line (SEF 12x10). Not once in numerous comparisons over a few years have we seen a lick of difference between them, circling around the target and comparing the signals with both machines. Simply put, a machine can't see what the coil can't see separately.
Now, I don't argue any one approach (low or no iron rejection, faster recovery speed, or a sharp DD line) is better than the other, but I do now believe the end result, as far as what my own eyes and ears have shown me, is pretty much a wash on whichever path you take. A few years back I believed the complete reverse of that, in that the only true way to unmask was with faster recovery speed and low or no iron rejection.
This might be somewhat controversial to say, but let's add a little more fuel to that fire...Think about low iron rejection. Is it really allowing you to see a coin in iron better and thus sounding off to the coin only because the iron rejection has been zeroed out? Not IMO. Detection fields, far as I always read, stop and interact with the first (meaning shallowest) object in the field and it's game over after that. I don't really think in most instances the field is seeing the iron and coin at the same time, but rather low iron rejection is just decreasing the lag time between "rejection mode" of the iron to "acceptance mode" of the coin, and thus you are able to hear the coin where as otherwise it might be silent as the machine resets from rejection to accept in it's processing.
By using a quality coil with a sharp DD line you are then more cleanly going from "I see the nail" rejection mode to "I see the coin" acceptance mode. In a sense then, you've found another way to increase "recovery speed", or better put perhaps you've turned the page more cleanly from one (the nail) to the other (the coin) as the coil moves. By lowering iron rejection you are eliminating the lag time between reject and accept, or by using a sharp detection field you are in a sense doing the exact same thing.
From what I've read, it appears it is possible to gleam the ferrous (nail)/ non-ferrous (coin, tab, ring, etc) traits from a detection field in a mixed target "as one" signal. Minelab has appearantly figured out how to do this, and I would suspect is then the reason why they just built in fixed iron rejection on the BBS units, as lowering it wasn't so much of any issue anymore. That's what I suspect anyway, and the ability to pull coins out of nails does appear rather good to me.
But in terms of being able to tell a mixed non-ferrous/non-ferrous two target signal, far as I've read over the years that's an impossibility as to what a detection field can tell you. Only the ferrous/non-ferrous aspects are what detection fields appear to be able to tell a detector, from what I've read anyway, but would welcome any links to the contrary on both these points.
But whether either one is true or not as said above, the laws of physics the govern detection fields appear that both targets have to be at the same depth and so close to each other they'll priobably be touching or overlapping, because the field stops and warps around/interacts with the first metal object it hits from all I've ever heard on the subject. You can see this demonstrated to the extreme as a test article once showed, in that even something as small as a staple will mask a coin deeper below it in the field. And once the staple gets deep enough you no longer will even hear the iron or null from it, which means now you don't even have a clue that something was under the coil that might be masking something else. For that reason many believe (myself included) that there are still tons of silver coins out there left to be found at spots long since given up by hunters as "dead".
Even if the two targets are at the same depth, once they get a certain distance from each other in the horizontal aspect, the field is going to hit and only see one or the other. Think of it like a mouth to a flame. For that reason a sharp DD line is of primary important to me, and also because a shallower trash item can be well off to the side of the coil while a deeper coin is directly under it, and you'll never hear that coin. First metal object hit and it's lights out IMO.
Recovery speed IS an issue to me only if I'm in the mood to swing fast, but so long as I control my sweep speed it's far more important to me to have a sharp detection field. If you simply have to swing fast then yes, you need a fast machine. On the other hand, for my style of hunting these days at dead sites I prefer to work the coil slowly and sniff in and out of iron or other trash looking for any hints of a high tone. You just can't do that with some fast machines, or at least moving the coil too slow might cost you performance in depth or discrimination ability if the machine isn't built to be used slow.
Besides, I feel a slow recovery speed gives me the extra time to really drink in and take the best look I can at a target without the machine trying to reset on me or clip the audio short when I'm after the finer traits of the target. To me this also becomes important when I'm trying to suck up the hardest hit on a target at fringe depth. I think of it as trying to read a street sign at 100mph vs only doing 35. I feel the longer hovering look at depth gets me that extra bit of depth to pull the best ID/audio I can out of the target. You can see this in play on some machines that have the ability to change the SAT (self adjusting threshold). Too fast a setting and depth suffers. Too slow and stability might suffer. In trying to be one or the other you often do neither well. Better to use a machine (fast or slow) that suits your style and stick with it.
Final thought, and part of your question I think- Do some machines have the ability to hold onto the right conductivity better of two non-ferrous mixed targets in the field (again, same depth then and very close for the perfect storm of masking)? Yes, I believe so. In truth in a mixed non-ferrous signal the machine doesn't know it's got two targets under the coil (it doesn't either with a nail and a coin, only that some machines can see the ferrous/non-ferrous aspects of the signal to some extent), and so what usually happens is the conductivity is summed between the two. Tab and silver dime= conductivity (ID) is placed on a average between the two (like say as a zinc penny then perhaps or a bit lower). But this averaging is also based on the size of both targets. Bigger piece of foil vs smaller silver dime= probably the ID shows much nearer to the foil range, like say in the nickel range or so.
But to answer your question- I've used machines that seemed better at holding onto the higher conductive properties of the silver dime and not allowing the VDI to be dragged down, or at least as much (to say a zinc penny signal perhaps), or at the very least as often as I've experience with prior machines. Why? You're guess is as good as mine. My only stab at a guess is this- As you wiggle the coil the field will still see one or the other separately here and there (with a sharp field anyway), and perhaps the VDI and tone processing of the machine tries to hang onto and offer up the highest conductive signal it saw in those split micro seconds between seeing one or the other, and then continues to show that as both are then washed in the field most of the time?
This concept can be seen in volt meters. Some have split second display stabilization where a constant up/down voltage quickly pusling is ignored and instead the highest voltage reading is the only thing seen on the screen. Despite all that, the audio still knows. You might not see the finer change from one to the other thanks to the buffering of the VDI, but the audio on a good machine will still warble a bit and say "something ain't right here, you've got a less than pure signal that changes a bit under the coil."
All the above said, this is just my working theories with some reading and experience over the years in this hobby. I'm sure there are others who strongly differ with some or all of what I just said above, and I can fully understand, as it wasn't too long ago I would have been the one arguing with myself, in particular on the various paths to unmasking and how important each were in the final outcome. I would love to read any articles or links that offer differing points of view on all this, as I don't claim to have all the final answers.