Cardano Updates

Two fixes for goroutine deadlocks that silently killed the chainsync
pipeline, causing nodes to stop advancing with no error output.

1. Close result channel on reader exit: ledgerReadChain() could exit
   on errors (chain reconcile failure, block decode, nil iterator)
   without closing resultCh. The consumer blocked forever on the
   channel read. Add defer close(resultCh) so the consumer always
   unblocks and the caller can create a fresh reader.

2. Preserve done channel for cached batches: when an epoch rollover
   split a batch, the next loop iteration reset currentReadResultDone
   to nil before consuming the cached remainder. The reader goroutine
   blocked forever waiting for its done channel to close. Move the
   nil reset inside the else branch so it only runs when reading
   fresh from the channel.

Signed-off-by: wcatz <[email protected]>
Signed-off-by: wcatz <[email protected]>

Use safeAddUint64(K, K) instead of 2*K to prevent uint64 overflow
when securityParam is large.

Signed-off-by: wcatz <[email protected]>
Signed-off-by: wcatz <[email protected]>

Bumps [benchmark-action/github-action-benchmark](https://github.com/benchmark-action/github-action-benchmark) from 1.21.0 to 1.22.0.
- [Release notes](https://github.com/benchmark-action/github-action-benchmark/releases)
- [Changelog](https://github.com/benchmark-action/github-action-benchmark/blob/master/CHANGELOG.md)
- [Commits](https://github.com/benchmark-action/github-action-benchmark/compare/v1.21.0...v1.22.0)

---
updated-dependencies:
- dependency-name: benchmark-action/github-action-benchmark
  dependency-version: 1.22.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <[email protected]>

After a stall, recorded peer tips go stale while the network advances.
When a known peer's tip jumps by >K blocks (Case 1) or a new peer's
tip exceeds best_peer + K (Case 2), the check rejects the tip as
implausible. This creates a deadlock: the node can never advance
because it rejects all peers that have the blocks it needs.

Add a catch-up relaxation: when the reference block is at or behind
the local tip (indicating stale peer records), widen the acceptance
window to 2*K of the local tip. This breaks the deadlock while still
rejecting spoofed tips when the node is healthy (reference > local
tip means peer records are current).

Signed-off-by: wcatz <[email protected]>
Signed-off-by: wcatz <[email protected]>