regression: mismatched types

[Mark-Simulacrum]

https://crater-reports.s3.amazonaws.com/beta-1.43-1/beta-2020-03-12/gh/finegeometer.vector-spaces-rs/log.txt

It looks like type parameter order got switched somewhere somehow, though I am unclear how that managed to happen. Possibly related to recent inference work to make binary operators work for references?

[Mark-Simulacrum]

Also leaving https://crater-reports.s3.amazonaws.com/beta-1.43-1/beta-2020-03-12/reg/yass-aux-0.1.1/log.txt here in case it's the same underlying bug (looks quite similar at a high level) but we should file a separate issue if not.

[Mark-Simulacrum]

cc @Aaron1011 who I think has done some inference related work recently

[Centril]

Closing as expected fallout from #69340 and wontfix.

#70444 (comment) might be unrelated though.