I managed to get the other relay off the board without too much trouble, and without breaking the relay or board, and it also showed very different contact resistances for the two poles.
I decided after a while to carefully de-lid the relays and see what was going on. Here are some pictures of the innards of the relays. I had figured there may be a bad alignment or schmutz on the contacts, or whatever. Can't hurt to have a look!
So of course I ran my 11amp tester over and over on the relays while they were open, and found that the voltage drop(and inferred resistance) was all over the place. Even bought some new test leads because I thought my old ones had become unreliable! I tweaked the alignment just a bit, but things stayed way more variable than I'd expect. Eventually, I took a better look at the spring-steel of the contacts and realized they were discolored not at the >contacts< but rather up at the top, where the spring steel is riveted to the other PCB connection! (see the last pictures), immediately that pointed me to the issue. The rivets hadn't been fully squished at the factory, and as I was tweaking the alignment or moving the relay in any way, the resistance at those rivets would change(note that the 'top' picture is actually AFTER I squished the rivets better). Notably, squishing the rivets completely closed the contacts even without applying power to the coil, so it was clear they were loose(r) before squishing.
As I was experimenting, I found:
Simply turning the coil off and on a few times(without load on the contacts, like a tesla would have) would make the resistance change a bit randomly.
As the contacts were warming up, they would decrease their resistance a bit, to a point.
The alignment and angle of the contacts was not anywhere near as important as I expected. The 2 milliohm resistance contact is visibly cockeyed(not flat-to-flat), but I didn't want to screw with it.
After squishing and readjusting the contacts, I rechecked resistances and found that one relay had a pretty closely matched 6 milliohm resistance for each contact, and the other had 2 milliohm and 6 milliohm for its two contact resistances. Doing the math, that got me to about 6 watts total dissipation for the first relay, and 4 watts for the other relay(both at 48 amps, apportioned appropriately for the contact resistances).
I was very satisfied with these results, of course, I re-lidded both relays with high-temp epoxy, reattached them to the board(adding a bit less than 1/16th inch spacing off the board for better airflow, and nipping off the excess wire of the optical temperature sensor that extended through the board). I was careful to put the 'better' of the two relays over the temperature sensor, but I really don't think it matters.
I reassembled the entire thing(no hacky heatsinks!), and she lit right up and gave Stacy's Mom some electrons! Sadly, I had Stacy's Mom to only accept 24 amps from the last time I tried this charger, but figured that out after a few hours of grumbling over what could be wrong, and taking it back apart. Finally set Stacy's Mom to 48 amps, and it ran right up there. Since the covers were off now, I could feel that the relays were only slightly warmer than ambient temperature(60f) after a half hour of 48 amps. Additionally, the voltage losses from in-to-out on each leg were 0.08 volts, much better than the prior 0.21 and 0.285, respectively.
I've now put all the covers back on and fully expect my next infinite hour charge session to go just fine!