Add proof types for synthetic porep (FIP-0059)

[cryptonemo]

@Stebalien Does this help as an alternative to #1335?

[cryptonemo]

Alternatively, after your branch is merged, I could rebase against master. Let me know what makes sense.

[Stebalien]

I was going to do it in two PRs.

[cryptonemo]

Ok, that works, thanks

[Stebalien]

This looks correct, but I think this'll need to be tested in a devnet before we can actually be sure we haven't missed something.

[codecov-commenter]

Codecov Report

Merging #1409 (81b39e5) into master (2b1df06) will decrease coverage by 0.07%.
The diff coverage is 89.47%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1409      +/-   ##
==========================================
- Coverage   91.07%   91.01%   -0.07%     
==========================================
  Files         145      145              
  Lines       27397    27409      +12     
==========================================
- Hits        24951    24945       -6     
- Misses       2446     2464      +18     

Files Changed	Coverage Δ
actors/miner/src/commd.rs	82.85% <60.00%> (ø)
actors/miner/src/policy.rs	92.00% <100.00%> (ø)
runtime/src/runtime/policy.rs	98.51% <100.00%> (+0.14%)	⬆️

... and 3 files with indirect coverage changes

[anorth]

I don't think we want to change any existing tests here. Possibly we need some new tests, but if this value is just plumbed directly to the syscall, we're probably ok without.

[cryptonemo]

Thank you! If we do end up needing a separate test, let's figure out the best way to do that.

[anorth]

@Stebalien everything needs testing in devnet, right? I don't think that should hold us off from merging code towards FIP-approved changes into master. Integration will be difficult enough regardless.

[Stebalien]

Yeah, that's fine by me. I'd say that this change is especially hard to get confidence on without that, but there's no harm in merging and testing later.

[rjan90]

Fwiw, we have done a couple of rounds of testing (local + butterfly) with a locally built actors bundle based of https://github.com/filecoin-project/builtin-actors/tree/dev/synth-porep-1 in Lotus.

Plan on the Lotus side is to do a round of SynthPoRep-regression test in a devnet, and also reset butterfly next week-ish. It would be really useful to have these changes in "tag/release", so we can get a bundle as an artifact, instead of building it locally.

[cryptonemo]

Rebased against master

[anorth]

@Stebalien this LGTM, could you please merge after verifying for yourself?

[Stebalien]

@rjan90's testing is enough for me and I don't have a setup where I could verify this.