Do you even use GD&T position tolerance with MMC on these holes? If so I can't see how your idea will work. That said you only mention +- so I guess maybe you don't.
I'd suggest an inseparable assy where you just worried about the location tol of the parts being soldered in but your customer requirements seem to prevent this, or at least complicate it.
Drawoh, I've seen the note "Dimensions Apply After Plating" and it's in the standard. I have no problem with this and in most cases believe it is the way to go. However, the idea of splitting location and size tolerances so that one applies before and one applies after plating sounds questionable. I have never seen this.
OP - Why not just have loser location tolerances on the holes to accommodate the plating shift, assuming function allows this, and make the tolerance apply after plating. This is effectively what you're doing anyway, you're adding the potential .003 to you location tolerance.
Maybe you could even take advantage of what ASME Y14.5M-1994 1.4 f says about adding non mandatory process dimensions, although in this case it would be a 'process tolerance'. Put a tighter tolerance for before plating qualified by the "NON MANDATORY" note. I believe this would be subtly different from a ref dimension which wouldn't have a tol.
If you aren't using GD&T position for these holes I'd seriously consider it as in most cases it will actually allow more usable tolerance for the same functional requirements.
Posting guidelines faq731-376 (probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484