Attempts to make tests work better using the buildkite plugin

[jagoosw]

Not making this a draft or the CI won't run

[jagoosw]

So this may be working now, so I'm going to make a branch from this branch and PR it into main, then make them both run at the same time to see if that works as a stress test

[jagoosw]

This is way way slower for some reason? In #186 (old version of pipeline) the tests finish in <20min but this is still running after 45min.

[jagoosw]

Maybe I should experiment with what we're caching

[jagoosw]

Ah it realises that these branches are duplicates I think

[jagoosw]

This does seem a lot more robust now but also a lot slower, I will investigate adding more caching

[jagoosw]

I don't think we should do it this way since its so slow