add MSM builtin

[perturbing]

Implement the bls12_381_{G1/G2}_multiScalarMul builtin for Plutus-Core and Plutus-Tx.

The current implementation depends on a SRP of a PR against cardano-base (where the C bindings are added for the underlying functionality).

[kwxm]

This looks basically OK, but I think we should change the inteface to elimitate pairs in PLC. You could also add changelog entries in plutus-core and plutus-tx saying that you've added preliminary versions of the MSM functions.

I think it's probably OK to merge this with the SRP in cabal.project, but your changes will definitely have to be merged with cardano-base before, eg, a node can be released that uses a Plutus release involving the SRP. @zliu41: is that reasonable?

[perturbing]

@kwxm, the manual on adding a built-in clearly states

PLEASE ADD ALL NEW CODE AFTER THE CODE FOR EXISTING BUILTINS SO THAT CONSTRUCTORS AND FUNCTION CLAUSES APPEAR IN THE HISTORICAL ORDER IN WHICH THEY WERE IMPLEMENTED.

Should I also follow this for plutus-tx and plutus-tx-plugin etc?

[kwxm]

@kwxm, the manual on adding a built-in clearly states

PLEASE ADD ALL NEW CODE AFTER THE CODE FOR EXISTING BUILTINS SO THAT CONSTRUCTORS AND FUNCTION CLAUSES APPEAR IN THE HISTORICAL ORDER IN WHICH THEY WERE IMPLEMENTED.

Should I also follow this for plutus-tx and plutus-tx-plugin etc?

Actually I don't know. The thing about adding stuff at the end was mostly for plutus-core. It looks as if we've mostly been doing the same thing in plutus-tx, but now I'm wondering if the Haddock will be confusing if we have new functions that are separate from the previous ones. I think we have some generated Haddock somewhere so I'll try to find that and see what it looks like.

[kwxm]

The Haddock for PlutusTx.Builtins is here and that does have related functions grouped together, so in plutus-tx and plutus-tx-plugin you should probably put the new functions in the same place as the existing ones.

Edit: no, wait. It looks as if the Haddock generator is grouping things together itself. I'm not sure how it does that. Anyway, just stick the new BLS functions in the same place as the old ones.

[perturbing]

@kwxm, for now, this PR only adds to plutus-tx and plutus-core, for which both test suites

• cabal test test:plutus-core-test
• cabal test test:plutus-tx-test

pass (I updated the interface). That said, this PR does break other packages (plutus-tx-plugin for example), what is generally the approach here? This steps suggest that this now can merge, but this will break main?

The new builtin should now automatically become available in Plutus Core. At this point it is safe to merge the code, although the new builtin will be prohibitively expensive for real-world usage.