Replace double core CE with two sequential single-core CEs

[ilyamandel]

Currently, we always assume that if both stars are evolved when CE commences, a double core CE should unbind both envelopes; in reality, if stars are in very different evolutionary states, one may have a tighter-bound envelope that may remain bound throughout the CE event (i.e., the secondary may essentially be a point particle for the purpose of the CE resolution).

I propose that we always treat the CE first as a single CE, whatever the state of the secondary, using only the binding energy of the donor to determine the final separation. Then we check if at the end of the CE, the secondary overflows its Roche Lobe. If it does and has an envelope, we immediate go into a second CE phase, this time originating from the secondary. If the secondary has no envelope, and either the secondary or the primary overflows the Roche lobe immediately after CE, we put up our usual RLOF-immediately-after-CE flag and effectively treat this as a merger. Otherwise, the system survives CE without doing anything to the secondary’s tightly bound envelope.

An alternative (proposed by Selma) is to check for stability of MT if there's RLOF at the end of CE. Personally, I suspect that in this case, there will be significant interaction between the two envelopes during the CE, i.e., this is really a double-core CE case -- either both envelopes can be ejected, or there's a merger.

[ilyamandel]

I think this is an enhancement worth doing, albeit with a reduced urgency, so will leave it alive for now.