TST: downstream instability

[neutrinoceros]

Description

looks like merging #200 caused a regression in yt
https://github.com/yt-project/yt/actions/runs/3223984952/jobs/5274634351

I'll take care of this today if I can

[neutrinoceros]

So the problem is that, in yt, some code has been written that relies on the old behaviour of, say np.dot, dropping units on unyt arrays.

The code can (and likely should) be adapted on yt's side, but it seems to me that it'd be best to consider that a fully NEP 18 compliant version of unyt (as is being developed starting with #200, see #289), should be numbered 3.0.0 instead of 2.10.0.
@ngoldbaum WDYT ?

[ngoldbaum]

Yes, definitely, the next release is unyt 3.0

[neutrinoceros]

Cool. I don't think there's any actionable left for now so I'll close this but just to make sure everybody is on the same page I'll ping @jzuhone