l4v proof testboard should filter by architecture about ci-actions

[axel-h]

I wonder if trying to automatically detect the targets that are impacted by a change is possible? Using manual tags might not catch the subtle things. There is a price because some source or binary diff requires assembling the thing to diff - but that might be cheap compared to the proof run.

[lsf37]

Ah, I should have been more precise. Context is difficult :-) This issue is for a different testboard -- it's for when you have changed the proof and want to do a test run on your changes before you make a PR. This is esp useful if you're working from a fairly limited machine.

For testing C changes, I agree it's not a good idea to test these manually. We could try to figure out from preprocess results which images need testing, but it's often enough all or none that it doesn't matter so much in terms of VM limits. (Whereas for proof changes you might want to run a few of these)

[lsf37]

I'm closing this, because with the AWS PR tests we now have, the testboard has become essentially unused.