replay: refactor set-root to enable alpenglow to take over

[AshwinSekar]

Problem

In alpenglow, replay will no longer be the thread setting root. To prepare for this we refactor the set root to the votor crate.

Summary of Changes

This pulls in the votor crate from the Alpenglow repo but only for the root_utils file. The full upstreaming of this crate will take place after VoteStateV4 has landed in sdk master (bunch of dependencies to resolve that need this 😅 ).

The callgraph is

///
/// ReplayStage::check_and_handle_new_root -> root_utils::check_and_handle_new_root(callback)
///                                                             |
///                                                             v
/// ReplayStage::handle_new_root           -> root_utils::set_bank_forks_root(callback) -> callback()

This extra callback will take care of rooting the TowerBFT specific structures to maintain compatibility. In the future Alpenglow will call the root_utils function without the callback

[codecov-commenter]

Codecov Report

❌ Patch coverage is 93.18182% with 12 lines in your changes missing coverage. Please review.
✅ Project coverage is 83.2%. Comparing base (f6e6ff2) to head (189593b).
⚠️ Report is 4 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff            @@
##           master    #7452     +/-   ##
=========================================
- Coverage    83.2%    83.2%   -0.1%     
=========================================
  Files         800      801      +1     
  Lines      362783   362869     +86     
=========================================
+ Hits       302195   302239     +44     
- Misses      60588    60630     +42     

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[bw-solana]

Out of paranoia, I inlined all of this and plugged into a diff checker. Code looks identical except for the following:

1. We pass in parent_slot instead of vote_bank <-- seems like a good change
2. We pass in highest_super_majority_root instead of block_commitment_cache <-- again, seems good
3. Reverting to positional formatting of arguments for err <-- I think we should go back to string capture style
4. Printing out pubkey when setting root <-- we pass pubkey through a couple API layers just for this.. is it worth it?

[AshwinSekar]

Reverted 3 189593b

As for 4, it makes debugging local-cluster tests easier. Perhaps a bit overboard but in alpenglow i've made every log print pubkey so that we can see what's going on.
If you prefer it removed I can keep a log w/ pubkey in event_handler when we call set root instead.

[bw-solana]

It feels weird tunneling down through a couple API layers just for local cluster test, but I agree it's soooo valuable to have for understanding those tests. I'm fine with it

[AshwinSekar]

agreed, would be nice if we could have some log magic to tag on pubkey in local cluster haha

[steviez]

I was out today and saw @bw-solana already approved - can you give me the chance to review tomorrow (Tue) before merging