CI: unpin numpy for CI / Checks github action

[simonjayhawkins]

xref #39513

[WillAyd]

Cool thanks for taking a look. Are you expecting this new environment to be part of CI or to just leave here for visibility as we clean up the existing failures?

[simonjayhawkins]

Cool thanks for taking a look. Are you expecting this new environment to be part of CI or to just leave here for visibility as we clean up the existing failures?

open to suggestions.

vendoring stub or adding additional steps to current check (update environment and run mypy again) are other options.

[jbrockmendel]

@simonjayhawkins is getting this passing viable? a bunch of the complaints are about e.g. np.random.randn, not clear how to resolve

[pep8speaks]

Hello @simonjayhawkins! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2020-10-03 15:46:30 UTC

[simonjayhawkins]

a bunch of the complaints are about e.g. np.random.randn, not clear how to resolve

probably need to submit some fixes upstream

is getting this passing viable?

imo not with the current review process, even though I normally put the errors in the OP for typing fixes

I'm now going to add the errors directly to code here, so it that it may be easier to identify the problems (and the low hanging fruit)

I may open another PR that sits on top of this with the fixes so that it is easier to see the reason for the code changes from the diff

(if this process is more successful, could maybe do the same for check-untyped-defs, many fixes were sorted 12 months ago, trying to get them into the codebase is a nightmare)

[jreback]

this lgtm @simonjayhawkins rebase once more just to be sure and let's merge!

[simonjayhawkins]

ok. will merge shortly if green (hopefully will be, one 1 commit to master since last rebase)

There will likely be some ci errors with merging subsequent PRs that are not rebased to this one. I think that we could take the hit and follow-up with ci fixes rather than asking too many PR authors to rebase.

[jreback]

thanks @simonjayhawkins very nice!

[jreback]

ok. will merge shortly if green (hopefully will be, one 1 commit to master since last rebase)

There will likely be some ci errors with merging subsequent PRs that are not rebased to this one. I think that we could take the hit and follow-up with ci fixes rather than asking too many PR authors to rebase.

yep we'll see and that's fine too