fix: allow secp256k1-recover? to accept high-S signatures

[brice-stacks]

Fixes discrepancy in updated secp256k1 implementation with old version.

[jcnelson]

We sure this fixes the chain sync test?

[brice-stacks]

We sure this fixes the chain sync test?

It definitely fixes the one I investigated, https://explorer.hiro.so/txid/0x86ef14fa87fb7de8aef4955290637a7b216d8a47378b7890a87065e41cc8e09c?chain=mainnet

[brice-stacks]

For this release (3.3.0.0.0) we are going to revert back to the old library for secp256k1 (#6645). We will come back to this fix with more time for proper testing.

[jcnelson]

@brice-stacks Since the reversion to libsecp256k1 is complete, can this be closed?

[brice-stacks]

I think we may still want to switch to the new crate since it is nice that it is all Rust and it removes the need for the separate implementation for Wasm. We just need the time to properly test it.

I can check and make sure that this draft implementation doesn't have the same discrepancy that the r1 implementation had.

[brice-stacks]

I confirmed that for secp256k1 here, I did use the correct _prehash variants, avoiding the double hash situation that we saw in secp256r1.