Feat/block rejections timeout heuristic

[rdeioris]

Description

This patch modifies the logic of miner timeout while waiting for signer confirmations/rejections.

The current behaviour is to wait indefinitely until the threshold is reached (for both rejections and confirmations) or the burnchain tip changes or the block is found in the staging db.

The new behaviour adds heuristic for introducing a timeout (instead of waiting indefinitely and in addition to the threshold checks) that decreases based on the amount of rejections.

Note that if the initial timeout (the one set at the start of the cycle) expires without having rejections, the old behaviour of retrying from scratch is still valid. (eventually the other mechanisms, like the change of the burnchain tip will kick in as before)

The timeout steps can be configured from the config file:

[miner]
block_rejection_timeout_steps = { 0 = 100, 15 = 200 }

That configure the system to wait 100 seconds when there are 0% rejections and 200 when there are 15% rejections. The user is free to specifies more steps:

The default is:

(0, Duration::from_secs(600));
(10, Duration::from_secs(300));
(20, Duration::from_secs(150));
(30, Duration::from_secs(0));

Applicable issues

• fixes Miner should abort waiting for proposals #5705

Additional info (benefits, drawbacks, caveats)

Note that the original code was set to trigger the conditional variable of the waiting cycle only when reaching the threshold (in confirmations or rejections). Now it is triggered at every update to move the logic of timeout in the block proposal final part ( get_block_status() )

The integration test is based on a test only atomic counter (added to the Counters struct) that gets updated with computed timeout value based on rejections.

Checklist

• Test coverage for new or modified code paths
• Changelog is updated
• Required documentation changes (e.g., docs/rpc/openapi.yaml and rpc-endpoints.md for v2 endpoints, event-dispatcher.md for new events)
• New clarity functions have corresponding PR in clarity-benchmarking repo
• New integration test(s) added to bitcoin-tests.yml

[rdeioris]

@obycode I have refactored the logic to keep the old behaviour and just check for change in the rejections amount. Combined with timeout checks i got no more stalls while waiting for the condvar

[obycode]

Thanks!

[hstove]

Mainly requesting a change to the integration test, because you can simplify the code a lot.

Nice work on this! The miner-related code is very straightforward.

[obycode]

LGTM, just had a minor comment to help remove some type casting.

[obycode]

👍