a really interesting study - and you must be going nuts comparing multiple sources!
My initial thought, for what its worth, is to suggest great caution in trying to extrapolate unit values from analysis of those who left. For example, it may be (a hypothesis not supported by any evidence to hand) that a higher proportion of 'leavers' came from areas traditionally royalist in their sympathies - Bouches-de-Rhône, the Var and the Vendée spring to mind - and left for reasons of pragmatic assessments of where the future lay. Others may have left through simple campaign fatigue and disappointment that the last throw of the die had failed. Yet others may have stayed on in whatever units they were then transferred to because they knew no other method of making a living...or decided that the benefit of being a grognard (versus a grunt) may justify staying in to see what happened under the Restoration. It's difficult to prove any of this, of course, without eyewitness or contemporary accounts. But a fascinating task, I should think!