Many brewers, home and professional, reserve dark grains until sparge or kettle but most seem to do it to keep the harsher flavor elements from these grains out of their beers rather than to simplify pH management though it does do that through the mechanism that you suggest. There are a couple of things to consider if you do this.
1. The beer won't taste the same. Roast barley steeped in the kettle or sparge water won't impart the flavor the roast barley included in the mash does. This is exactly why Gordon Strong and one of the head brewers at a local Gordon Biersch keep it out of the mash. But you
may (note that this is in italics) lose some flavor components you want by doing this. Less extraction of acrid flavors, sharp flavors but less extraction of coffee flavor too.
2. You keep the acid of those colored malts out of the mash thus making the pH of the base malt the main determinant of the malt component of mash pH proton deficit. But that acid still goes into the kettle and its effect on kettle pH needs to be considered or at least kettle pH needs to be determined. This would especially be the case if, as would be typical for an Irish stout brewed with water of typical municipal supply alkalinity the proton surfeit of the roast barely were barely sufficient or even somewhat insufficient to bring the mash pH into the desired range necessitating the use of another acid source for that purpose (we'll look at that example in a minute)
3. It is no harder to determine the proton deficit (or surfeit) of a specialty malt than it is of a base malt. If we ever get to the point where maltsters supply that information then all you have to do is punch that data into a spread sheet or calculator.
In a recent post,
https://www.homebrewtalk.com/f128/mash-ph-prediction-control-442357/, in which thread this discussion should probably be taking place, I describe a spreadsheet which can be used for pH prediction. It is, perhaps, as valuable for the insight it lends into problems like this one as it might be for pH prediction. When I make Irish Stout I mash 60 pounds of MO with 8 lbs roast barley and 13 of flaked barley in 25 gal of water with about 1.5 mVal alkalinity (75 ppm as CaCO3) and a like amount of total hardness more or less evenly split between calcium and magnesium. Sticking those numbers in with some assumed parameters for the malts/barley derived from my own measurements on Weyermann Pils and Kai Troester's measurments on the others (with a WAG on the flaked barley) I get a predicted mash tun pH of 5.60 and a measured mash pH of 5.57 with the roast barley in the main mash. The spreadsheet reveals that the roast barley is contributing 129 mEq of protons and that it takes 114 of those to move the water to 5.6. Were I to hold out the barley I would need to get that 129 mEq of acid from elsewhere (129 mL of 1 N acid) so that when I added the barley later I would have 129 mEq more acid than I would otherwise.
In another recent example I looked at based on a question here where RO water was involved (0 alkalinity) that the colored malts' proton surfeits were insignificant compared to the base malt's deficit which had to be made up largely with lactic or other acid. This would be the case with most lighter colored beers.
You may wish to play around with this spreadsheet a bit yourself to try to gain similar insight. The thread referenced above contains links to it and to a set of power point slides which sort of describe it but are not really a user's manual.