Incorrect orbital evolution on mass loss

[ilyamandel]

When a binary (no tides, no GWs, no supernovae, no mass transfer) loses mass in winds or pulsations on timescales much longer than the orbital timescale, M_total * a should be conserved. I've tested this before, but now it's not conserved, again. :(

The following example was created with ./COMPAS -n 1 --detailed-output --random-seed 1733362410

Part of the issue is that the binary evolution on wind mass loss is unnecessarily approximate, yielding wiggles (not visible on the plot without zooming in). I fixed that in #1304 (draft for now). But there's clearly a bigger issue when the star loses a lot of mass in pulsations on the TPAGB, leading to full envelope loss. To be fixed [I'll work on this, just listing @SimonStevenson , @reinhold-willcox , @jeffriley FYI .]

[ilyamandel]

Fix completed, maximum fractional change in a*M is now 0.0few percent (see plot below).

The main issue was that we would sometimes create a WD as an outcome of a supernova: if the Hurley criteria called for an ECSN, but we didn't meet the mass transfer history threshold for an ECSN proposed by @reinhold-willcox , ResolveElectronCaptureSupernova() would yield an ONeWD. That is inconsistent with the Hurley evolution in the HR diagram, and is also inconsistent with the idea that a WD is formed by gradual mass loss, not a rapid, explosive SN. For now, the fix is that if the ECSN criteria are not met, we go through a "standard" SN.

[ilyamandel]