Stabilize CI lint shard timing out

[Eric-Arellano]

Now that we're using both V2 isort and docformatter, we've seen most CI runs time out for the lint shard. This is because we do not cache the V2 results and the V2 linters have atrocious performance when the cache is cold, per https://docs.google.com/document/d/1Tdof6jx9aVaOGeIQeI9Gn-8x7nd6LfjnJ0QrbLFBbgc/edit#.

A followup could use remote execution with the lint shard.

[jsirois]

Excellent. I was hitting this in a PR and was assuming it was my fault. If this doesn't consistently solve it would be wonderful to revert the docformatter use until it can be made performant. Master and all PRs are effectively red otherwise.

[Eric-Arellano]

I was hitting this in a PR and was assuming it was my fault.

Me too.

If this doesn't consistently solve it would be wonderful to revert the docformatter use until it can be made performant.

I think I'll pursue remote caching later today, as V2 linter caching is no longer my #1 priority so we won't get a more sustainable fix for the performance quite yet.