test: niporep: e2e testing

[rvagg]

Replaces #12045

Initial form here has e2e onboarding of a single sector. Need to add:

• >1 sector in batch (note that real-proofs is quite expensive for this, so >1 sector might be slooow in CI)
• Check sector-activated event(s) (NOTE: needs fixing in builtin-actors: https://github.com/filecoin-project/builtin-actors/pull/1537/files/#r1648458930)
• Snap piece(s) into onboarded sector (and check events)
• Error: 0 sectors
• Error: >65 sectors (somehow .. maybe only with mock proofs)
• Bad proof type, bad aggregate type (maybe)
• One bad proof, one good (maybe, somehow, see Non Interactive Proof of Replication builtin-actors#1537 (review))
• Bad sealer id, bad sealing number
• Bad seal rand epoch (can't doo historical, we have a genesis chain and earliest is 180 days, but we can do future)
• Bad proving deadline (current or next)

This branch uses filecoin-project/builtin-actors#1537 for the actors bundle, but it also has filecoin-project/builtin-actors#1553 applied on top of it because otherwise preseals don't onboard and we can't get the chain started. It also has filecoin-project/filecoin-ffi#460 applied for filecoin-ffi because we need verify-signature.

[github-actions]

All checks have passed

[rvagg]

• Changed base to chore: test: refactor, fix and improve unmanaged miner #12143 until that's merged
• Rebased out the remaining pieces that have now been separately merged in to master
• Included https://github.com/filecoin-project/builtin-actors/releases/tag/v14.0.0-rc.1 bundle in here
• Squashed down to 2 commits - niporep test and actors bundle, nice and simple

[rvagg]

• Added LOTUS_RUN_VERY_EXPENSIVE_TESTS and put niporep w/ real-proofs in that category. It won't run in CI because it can't. We need to work on that. See [DX Streamline] LOTUS_RUN_EXPENSIVE_TESTS periodically outside of PR workflow #12136
• I'm cutting scope on this particular PR to get it landed. What I want to add is multi-sector onboarding, currently in both mock and real proofs we only onboard a single sector and then snap a deal into it. I think I'd like to onboard 3 sectors and have one fail but the others pass. Or at least do 2. Or some combination that makes sense and doesn't make real-proofs take too long.

Ready for review

[aarshkshah1992]

We can get the test for >1 sectors in a separate PR.