apollo_mempool: add timestamp logic to flow

[ayeletstarkware]

Note

Medium Risk
Changes transaction selection semantics in Echonet mode by gating FIFO pops on timestamp thresholds, which can affect batching/order and stall processing if timestamps are missing or inconsistent; other modes are largely unaffected.

Overview
Adds an update_timestamps flow from the HTTP server (POST /gateway/update_timestamps) through GatewayRequest::UpdateTimestamps into the mempool, allowing external timestamp mappings (tx_hash -> UnixTimestamp) to be applied.

Refactors the Echonet FIFO queue to be hash-based and timestamp-aware: get_timestamp() now derives a batching threshold from the first queued tx’s timestamp (and reuses the last value when empty), and get_txs() only pops transactions matching that threshold; tests/utilities are updated accordingly and Grafana dashboards gain client/server latency panels for the new request variant.

^{Written by Cursor Bugbot for commit 4ba2f2d. This will update automatically on new commits. Configure here.}

[reviewable-StarkWare]

This change is 

[ayeletstarkware]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• apollo_mempool: call echonet get_ts from communication add_tx #12741
• apollo_mempool: remove connection of gateway, http server from echonet #12740
• apollo_mempool: add timestamp logic to flow #12701 👈 (View in Graphite)
• apollo_mempool: implement fifo queue #12672
• apollo_mempool: add deployment mode config #12670 : 1 other dependent PR (#12850 )
• apollo_mempool: add queue trait abstraction #12660
• apollo_consensus: Add get_ts for Echonet #12535
• main-v0.14.2

This stack of pull requests is managed by Graphite. Learn more about stacking.

[cursor]

Missing body size limit on update_timestamps endpoint

Medium Severity

The /gateway/update_timestamps route uses post(update_timestamps) directly, unlike all other POST routes which use self.post_method_router(...). The post_method_router applies a DefaultBodyLimit::max(max_request_body_size) layer. The new endpoint bypasses this body size limit, allowing arbitrarily large request bodies that could lead to memory exhaustion.

 

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

[cursor]

Timestamp endpoint can panic on join

Medium Severity

update_timestamps wraps the call in tokio::spawn(...).await.expect(...). If the spawned task panics or the runtime is shutting down, the .expect panics the HTTP server handler, potentially taking down request handling instead of returning an error.