Fix Sequencer Proof Support

[preston-evans98]

Currently, state root assertions are disabled in test_proof_generation_doesnt_break_sequencer. To re-enable those assertions, we need to fix the following issues:

1. Proofs are excluded from replay planning.
   executor_events.rs:289 explicitly ignores ReadBlob::Proof in fetch_completed_blobs_by_sequence.
   This feeds resync/replay decisions via inner.rs:309, sync_state.rs:418, and sync_state.rs:467.
2. Proof events are no-op during catchup/replay.
   update_state.rs:349 handles DbEvent::ProofBlobAccepted(_) by doing nothing.
   That path is used in replay and final catchup loops from mod.rs update-state flow.
3. Proof ingestion path still has explicit TODO instead of executor processing.
   sync_state.rs:803 says TODO: Process proof blob in the executor; currently it only publishes side effects.
4. Replica path for proofs is unimplemented.
   replica/event_handler.rs:64 has DbData::NewProof { .. } => todo!().
5. Resync condition logic is batch-centric, so proof-only pending work is invisible.
   sync_state.rs:534 uses are_there_batches_to_replay.
   sync_state.rs:509 derives oldest unfinalized from batch list only.

[preston-evans98]

Blazej adds another item here: #2559 (comment)

This is not directly related to this PR but is relevant for the follow-up.

In replica_sync_task::run_handler, we handle two conditions:

DBDataRejected::ExecutorBehind – the replica executor is behind the DB updates (e.g., on startup when DA notifications are ahead). In this case, we wait until the replica catches up via DA updates.

DBDataRejected::ExecutorAhead – the replica executor is ahead, which can happen if DA is faster than DB notifications (e.g., during networking issues). In this case, the sync task skips DA events until they catch up with the executor.

The assert at replica_sync_task.rs:129 assumes the catch-up boundary event is BatchStart. After enabling proof processing, sequence numbers will also advance on proofs, so this case will need to be handled.

Note: This was handled by #2571

[preston-evans98]

This is pre-existing behavior, but it can became a real issue once we started publishing proofs through the blob sender.

Proofs can sit in the blob sender for longer than batches. A proof is published to DA right away, but it is only marked as processed once batch with higher seq nr was published (https://github.com/Sovereign-Labs/sovereign-sdk/blob/dev/crates/module-system/module-implementations/sov-blob-storage/src/capabilities.rs#L994). So if we have open batch 10 and proof 11, we need to wait for batch with seq_nr>11 to process the proof.

Because of that, proofs may stay pending for longer than batches normally do.
No need to do anything in this PR but we already discussed to increase blob_processing_timeout in the past, this will also fix this scenario.

#2571 (comment)