fix: ignore stale OutboundFailure events in bcast behaviour

[pinebit]

Summary

• Fix a bug where a stale OutboundFailure from a previously failed broadcast could kill a subsequent active broadcast
• When a broadcast fails (e.g. one peer times out), in-flight requests to other peers are still running. Their eventual failures would unconditionally call complete_active_broadcast, taking down whatever broadcast was active at that point
• The fix checks whether the op_id belongs to the current broadcast before acting. If it doesn't match, the event is logged and ignored — consistent with how SigResponse and MessageSent already handle stale events

Test plan

• cargo clippy --workspace --all-targets --all-features -- -D warnings passes
• cargo +nightly fmt --all --check passes
• cargo test --workspace --all-features passes

[emlautarom1]

@iamquang95 Any chance we could add a test for this edge case, or is it very complicated to reproduce?

[iamquang95]

@emlautarom1 let me try to add an unit test for this scenario