As far as mid process checking and temperature change. The tight holes I am checking are very small (3/32). I also assume I could just test with an eddy tester. I know everything plates differently based on geometry. All I really would want on the first part with an eddy tester would just assurance we are in a reasonable ballpark. I realize where my close tolerance areas were may be different and that could be documented for next time. Same thing with the witness sample method. Even if they were an approximation as reassurance you were in the ballpark of thickness I'd be satisfied.
In theory the witness sample should work, but it has to be the same stock of material for one, and racking has to be 'ideal' for both parts for them to behave the same way, what I mean by that is your large part may have surface area A, the small part might be A/10, and if you use, lets say 1 contact for the small part, you should use 10 contacts for the large part, to make sure the voltage potential on both parts is the same, if voltage is the same, then current density on both parts will be the same and that will ensure that both parts grow the oxide layer at the same rate
then there is the bath agitation aspect that Eaglemike mentioned, it must be sufficient so top and bottom portions of the tank are at the same temperature, if you want to test that, then you need to do the following, provide 2 sample parts to the anodizer, and ask them to rack one at the top and one near bottom of the rack in a batch of parts that will run the same current that your part(s) will, that way you will simulate the heat input into the tank and get an idea how top/bottom portions of the tank behave
regarding geometry and difference in plating, anodizing doesn't work like metal deposition plating does, where cathode is consumed to provide ions that then are deposited on the work piece, in that process, yes, geometry and cathode placement will affect plating thickness, in aluminum anodizing, the cathode is the acid, coating will grow evenly everywhere where the acid is in contact with the part - if the agitation in the tank is lacking, you might get temperature differential for one, second - small bubbles form on the surface of the part, which prevent acid from being in contact with the part thus retarding the oxide formation, those gas bubbles also can sometimes be trapped under a pocket feature of a part and prevent oxide formation, same thing happens in small holes - you can have acid in there at the beginning of the process, but as the gas forms inside that tiny hole, it displaces the acid and because of the geometry, the flow around the part can't flush that gas out - so the oxide on the outside of the part will be one thickness, while the hole may have almost no oxide thickness - happens mostly with blind holes, even worse if they are tapped, and even worse if the part has to then dyed black and the acid leaches out and eats the dye away... I often get to do parts with threaded blind holes right down to M1.6 size, I even try to rack them with those holes facing downwards, so gas pushes the acid out (customer doesn't request those threads plated), flush them by hand, and even then sometimes there are dye defects around the hole, always a lottery with those
so if your 3/32" holes are deep and blind and have to have the same oxide thickness as the outside of the part, you might be in trouble there...
anyway, I can't think of other issues you might run into, make a list, talk to the anodizers, try to talk with the operators, technical guys, not the sales people, those will always assure you that they have done thousands of parts like that with no problems etc.
p.s. noticed you said they are though holes, you might still want to "overplate" them and then run a diamond lap in there to get the final size for sure, little extra work for the peace of mind, might want to test that out with the test coupons mentioned earlier