Ed25519 async batch verification for JoinSplit signatures

[dconnolly]

Motivation

We've been verifying JoinSplitSigs one-by-one pre-ZIP-215. Now as we're post-ZIP-215,
we can take advantage of the batch math to validate this signatures.

Solution

This adds another tower::Service that allows async batch validation of Ed25519 signatures. This is very similar boilerplate to redjubjub::Verifier and similar tests.
I removed the single check and replaced its invocation with a call to this communal ed25519::Verifier instance.

I would have pumped all the joinsplits in our MAINNET_BLOCKS test vectors but these
signatures are over the sighash, which needs the NU code to compute, and once we're
doing all that set up, we're basically doing transaction validation, so.

The code in this pull request has:

• Documentation Comments
• Unit Tests and Property Tests

Review

Not urgent, a quick chunk of work we've recently been unblocked on doing.

Related Issues

Resolves #1944

Follow Up Work

This is the third async batch verifier tower::Service, and won't be the last (Orchard will bring halo2::Verifier and redpallas::Verifier eventually). There is a lot of boilerplate being copy-pasted around, which may be ok for easy deletion, but when it's duplicated amongst ~5 variants, it's a signal to revisit where we can DRY this up or make a more generic async batch Verifier, which can then be instantiated with different Items/ItemWrappers and some call method/function. Tracked in #1951.

These batch verifiers are returning their own errors from their underlying batchers, we should figure out the best way to turn those into TransactionErrors, either in zebra-consensus::transaction or in the async service layer.

[oxarbitrage]

I am not an expert on async code but i think this looks good.

I found a few minors that we can fix before merging:

• remove some not needed code and commented code.
• add "ed25519" string to some tracing messages that are too generic and can be confused with other similar ones.

[teor2345]

Looks great!

This PR could merge as-is, but let's do some tweaks and open some follow-ups first.

[teor2345]

I'm testing some full mainnet and testnet syncs locally with this branch and #1950.

I don't want to block this PR on those tests, it's not a high risk PR.

[teor2345]

I'm testing some full mainnet and testnet syncs locally with this branch and #1950.

Seems to work fine after 2 full mainnet syncs and 2 full testnet syncs, and after syncing to tip.

[teor2345]

Seeing 2 mainnet instances at around 180% CPU, spread across all cores.

(The CPU might even be a bit elevated due to all the "Invalid binding signature" logs.)

[teor2345]

Looks great, let's do more batching!

[teor2345]

I'm going to admin-merge because @oxarbitrage's changes were addressed.