I found the Victron RMA-self checklist, and did that:
https://www.victronenergy.com/media/pg/Pre-RMA_Bench_Test_Instructions/en/pre-rma-test-form---mppt-solar-charger.html#UUID-a02c41de-0c89-5d84-c701-4f28b6ae665b_bridgehead-idm133307965784142
Section 3 asks “Is the unit a 250/100 TR VE.Can model?” and allows me to send a warranty in purely on the rejection of that. Seems very generous, does not even get to the short testing. I wonder if that is a form flow mistake
Ignoring that, and continuing:
Section 4-1 diode between both positives: 0.5V either way.
in fact, all connections in all directions show about 0.5V, except between both negatives, which show about 1V. All resistances in all directions show OL.
Section 4-2 0.5A current limited 12V on bat: It shorts (to 0.5V), go to warranty
Section 4-3 0.5A current limited 12V on pv: It shorts (to 0.5V), go to warranty
There are some sub-clause conditions for those last two, but it should be noted that the install was ready and working and untouched for over a year. So that means I could not have reverse connected anything. Also these are a 4x 18V-6A panels in 2x2 setup, on a 100V30A device, so even in summer a panel short would not get near any 30Amp spec.
Section 4-3 does mean I have ridden around about a 10 days on shorted solar panels. However, with the mid-winter sun, I do not think that did much. Even right now, middle of the day, they measure 44V open, 0.8A short. At least they still work.
As can be expected from section 4.2, the external 30A fuse to the battery was blown. And this is why fuses are important. Though I suspect the internal 50A one would have blown otherwise.
PS: I realize no-one can really help. Just writing this down for myself really.
Still not sure what happened, but people on EEVblog seem to insist elco freezing was not likely the primary cause.