feat(zebra-state): Send a reference to latest non-finalized state in TipAction::Reset

[arya2]

Motivation

This is needed for the scanner task in zebra-scan (#7905) to get any blocks in the non-finalized portion of the best chain when there is a Reset action without repeated calls to the state service.

Possible Design

• Add the latest_non_finalized_state watch channel receiver
• Add a method on LatestTipChange that returns all the blocks in the latest non-finalized chain

This is acceptable because the worst-case scanning performance is 3 blocks per second, which means we can re-scan an entire non-finalized chain before the next block arrives. (Forks usually only happen every 300+ blocks.)

Alternative Solution

This solution is more complex than we need.

Too complicated

Add a method on LatestTipChange:

• accepting a previous block hash
• check if it matches any previous_block_hash in the non-finalized best chain (follow the old chain from the tip until reaching a common ancestor block)
• return all blocks after and including the block with a matching previous_block_hash, or
• return all blocks in the non-finalized best chain if it doesn't find the provided hash

Related Work

This ticket is required for:

• change(zebra-state): When the best chain changes, scan all the blocks in the best chain fork #7934

[teor2345]

I think we're missing the ticket that this ticket enables:

• change(zebra-state): When the best chain changes, scan all the blocks in the best chain fork #7934

[arya2]

I think we're missing the ticket that this ticket enables:

I was thinking this ticket could involve adding a method to be used instead of wait_for_tip_change() returning any new blocks on the chain that the receiver hasn't seen yet, so the task that scans new blocks as they come in would work for best chain changes too.

[teor2345]

@mpguerra did we decide to move this issue to a later sprint?

[teor2345]

I updated this ticket based on the current scanning speed. We don't need a complicated solution because it is fast enough:
#7813 (comment)