feat: allow for window post proving on a partition basis

[dignifiedquire]

This is based on #1497 and attempts to reduce the changes to a minimum needed.

TODOs

• improve naming
• improve documentation
• review for more things that can be simplified

[dignifiedquire]

@cryptonemo any thoughts why only darwin would show the testfailure?

[cryptonemo]

@cryptonemo any thoughts why only darwin would show the testfailure?

CI flakiness is my guess. The error is that it's missing parameters. I suspect a re-run may resolve it(?)

[dignifiedquire]

CI flakiness is my guess. The error is that it's missing parameters. I suspect a re-run may resolve it(?)

I am not sure, the same error happened on the original PR, but no other PRs

[vmx]

Rerunning CI should do the trick: cryptocorrosion/cryptocorrosion#48

[dignifiedquire]

didn’t work

[vmx]

It's a different error though.

[cryptonemo]

Very odd. And specifically because it's a window post test failing for code that's changed those parts, it's suspect. I'll see if I can find anything locally at some point. This is not the kind of code I'd normally quickly merge in any case, as everything touching PoSt needs careful review.

[cryptonemo]

As a first cut, this looks good. I was able to generate the parameters on darwin, so CI is good, but haven't been able to hook up the save/restore code, so it's done each time. It's not ideal, but if this can land in Proofs v10.1.0, it can be resolved after that.

[cryptonemo]

@cryptonemo any thoughts why only darwin would show the testfailure?

Resolved -- the test parameters are now generated on darwin.

[DrPeterVanNostrand]

Looks good to me, my only question is regarding the mixed use of get_partitions_for_window_post and get_num_partition_for_fallback_post.