Good information. So the battery acts as an electro-chemical clamp of sorts? Trying to think back to electrochemistry in college to understand what happens in that moment, but I suppose the voltage source + series resistance model works okay too. As I read in some other threads, some people disagree with the philosophy behind load dump protection, implying that "you have other more serious problems to worry about" if you get a 200V load dump spike on the DC bus.
It's my understanding load dump doesn't really...happen? With a battery? I guess the same physics should apply, the regulator can only slew so fast regardless, it should experience a swell from any load-drop condition, switching lights, motors, etc. Though maybe none of those loads are strong enough to overpower the battery, in terms of height or duration of that swell (it might be shorter than the full load dump case, even if it's the same time constants in play, which might also not be the case).
As for the battery itself, it behaves pretty much symmetrically in charge or discharge, up to the point where charging is complete, and where other reactions take over -- primarily water electrolysis. Electrolysis is a runaway condition, where due to bubble formation, the electrolyte cross-section is reduced, greatly increasing resistance, allowing terminal voltage to rise further. I would guess a car battery can do a good 20, maybe even 40 or 50A for brief charging in a low-state-of-charge condition (terminal voltage < 14.4V), but won't sustain such high currents for long; and in particular, as cells begin to fully charge (and this happens inhomogeneously across the plates: the facing surface has a direct line of sight; the pores behind, not so much), gas formation takes over and internal resistance rises. I would guess the current drops to ~20A steady-ish-state for terminal voltages in the 16-18V range. So there's a bit of a negative resistance characteristic, over a fairly long time scale (seconds).
As to philosophy, heh, for anything noncritical, indeed, who cares. The point of automotive tests is, everything built-in should last the lifetime of the vehicle. So it's worth testing even fairly rare events like this. But for like plugin aftermarket stuff, who cares, really. For hard-wired aftermarket stuff, well, I suppose that's up to the user, but one might argue the user isn't exactly inexperienced with replacing such things (obviously, it's been done once!). Or will necessarily be irate about random on-road failures, should it happen at a very inopportune time...
Speaking of aftermarket, it could well be that a replacement alternator might not be equipped with avalanche diodes, so the electronics should be protected even if everything is nominally to spec. One should hope such parts don't make it onto the approved replacement list, but who knows...
Tim