MARF Trie identifier refactoring

[kantai]

This PR mostly does two things:

1. Makes MARFs generic in the "trie identifier": allowing the burndb to use BurnchainHeaderHash directly with the MARF, rather than casting it into a BlockHeaderHash, and enabling the introduction of a new type StacksBlockId
2. Introduces a StacksBlockId type for capturing the index block hash. Everywhere where an index hash was previously used now uses a StacksBlockId (a lot of code changes!)

Those changes are are almost exclusively type refactors -- mostly no actually functionality has changed.

There was one instance, however, where a change was made -- in relay.rs, the MicroblocksData.index_anchor_block field was previously being filled with the anchored block hash, rather than the index block hash.

[lgalabru]

Looks good to me, thank you!

[lgalabru]

Should we also rename this block_hash instances?

[lgalabru]

Ha, isn’t “anchored block id” more accurate?

[jcnelson]

This LGTM, but I did have one lingering question -- besides unit tests, what's the point of keeping BlockHeaderHash around at this point?

[kantai]

This LGTM, but I did have one lingering question -- besides unit tests, what's the point of keeping BlockHeaderHash around at this point?

It only implements the MarfTrieIdentifier trait for unit tests (and is marked cfg(test)) -- I think it makes sense to leave that as is, since I think it's better to avoid changing old unit tests if possible.

[jcnelson]

Is there a difference between BlockHeaderHash-as-used-as-the-index-hash versus BlockHeaderHash-as-used-as-the-Stacks-block-hash? I can see that I'll need to replace the former usage with StacksBlockId in my PRs when this merges, but what about the latter?

[kantai]

I can see that I'll need to replace the former usage with StacksBlockId in my PRs when this merges, but what about the latter?

The latter isn't changed -- BlockHeaderHash is the type used for the hash of the Stacks block header (probably StacksHeaderHash would be the better name)