I suspect there are different Ag citrate complexes with varying amounts of light sensitivity. At least some of them are sensitive all the way down into green and possibly yellow.
Silver acts as a catalyst for decomposition of H2O2 into water and O2. The reaction is interesting because it proceeds via two different pathways through unstable silver oxides. If you make a test strip of fully developed silver ( "max black" ) and wash it carefully, and put it into H2O2, it will sit there happily bubbling away and do nothing. The silver is being constantly oxidized and then reduced with no net change. If there are any impurities ( e.g. Cl ) it will lighten as silver salts are formed. Both citric acid and sodium citrate will "stabilize" the H2O2 and slow down the catalytic decomposition, and if you use sodium citrate it will "bleach" to white, but the result is quite light sensitive. I've also tried NaCl and KCl in 3% H2O2 and both will completely rehalogenate in a minute or two... so the original image will re-develop, even if the bleaching is done in the dark ( and can make a nice warmer tone image, so this could be of interest outside of reversal processing )
H2O2 mostly leaves the AgBr alone, I'm not sure that EDTA or other choices will do that ( or pot-ferri ) we need it to be left mostly unchanged so that it can be used in the re-exposure.... I found a century-old reference on using H2O2 for silver bromide papers and it mentions that the H2O2 does not leave the AgBr completely unchanged... it makes it susceptible to physical development. If you mix one of the rehalogenating combinations, I think you can observe this: in the light tones, you can see darkening... this doesn't happen in areas unexposed to light and it's not visible in areas with more developed silver... on my test strips there are always one or two steps where this happens. I suspect this accounts for the "reversal" that Don sees. Anyway... all my attempts to regulate the activity of this bleach by adding Na-citrate ( to make a constant pH buffer ) have failed... some of them "bleach" very quickly, in less than one minute... but those are the ones that form very light sensitive Ag Citrate complexes. ( BTW Ag citrates are soluble in citric acid, but very little at the concentrations we are using ).
I'm a few hundred test strips into this, and still don't understand it....
If you slow down the decomposition of H2O2 too much, by adding too much citric acid... the bleaching does not stop, but it does change... it relies on dissolved air ( I think likely O2, but not sure )... which becomes locally exhausted very quickly and then the bleaching proceeds only as fast as the air can dissolve into solution, and if not agitated, faster near the surface. If you take a test strip like this out and let it sit on you desk and dry, it will bleach very quickly as the air becomes available. I think this explains part of the reason Joe finds different kinds of agitation work better. Right now... my working hypothesis is that our citric acid - H2O2 bleach is usually in some middle ground, where it is a combination of the reactions driven by H2O2 decomposition and dissolved air. The final result of these two are different: the dissolved air one leaves a cream or light tan "white" and it is more sensitive to light..... ( BTW it's important to note that BOTH are light sensitive, just that the Ag compound produced w/ only weak CA is much less so... this is why everyone is finding the amount of re-exposure matters ).
I'm still holding off writing more about my tests because I haven't started the final part yet. I've only been looking at bleaching and then how fast the bleached area prints out in light ( I did re-develop a few that included a halogen just to prove they act as expected, including a few produced with Na-citrate as the "pseudo-halogen". ). I suspect that the two forms of CA bleaching will re-develop differently because trying to make real photographs I've seen some come out with light tan highlights and others come out with nearly neutral highlights. I haven't proved it yet but I suspect what we'll find is that unfortunately these go in the wrong direction from each other: I think if we keep CA concentrations low enough that the rx is mainly driven by H2O2 decomposition, we'll get low sensitivity whites that develop into a rather ugly light tan color... and if we increase the CA until it's mostly dissolved gas driving the reaction, we'll get much more light-sensitive whites that take on a light-neutral tone depending on the amount of re-exposure. The good news is that at the end, if there is tan in the highlights, it can be bleached back to bright white in very weak pot ferri reducer...