evmonlyapp: time FinalizeBlock's stages around the executor

[bdchatham]

Describe your changes and provide context

On giga-testnet-2 the Autobahn main loop's execution share is ~81% of wall, but the executor's own phase timer (evmonly_block_phase_duration_seconds_total: open_view, occ_, encode_, commit_state, ...) accounts for only ~51%. The remaining ~30% is spent in FinalizeBlock outside ExecutePreparedBlock — and today nothing measures it.

This adds a PhaseTimer to evmOnlyApplication that charges every stage of FinalizeBlock to a phase, published as evmonly_finalize_phase_duration_seconds_total{phase=...}:

• take_senders — keccak of every raw tx + lookup of the senders CheckTx already recovered
• prepare — Executor.PrepareBlock: RLP decode + ECDSA recovery for every sender this validator did not admit (~39/40 of txs at 40 validators)
• execute — Executor.ExecutePreparedBlock (the executor timer breaks this one down further)
• tx_results — building the ABCI TxResults

Reset() at return closes the last phase, so the gap between blocks is charged to nothing. FinalizeBlock is serialized by the executor mutex, so a single timer is safe.

Measurement only, no behaviour change. It decides whether pipelining PrepareBlock for block n+1 while n executes (the second half of #4219's loadtest design, which the node does not do) is worth building: if prepare is the bulk of the unaccounted ~30%, it is.

Testing performed to validate your change

• New TestEVMOnlyApplicationTimesEveryFinalizeBlockPhase: binds the timer to a manual OTel reader and asserts all four phases are recorded by one FinalizeBlock.
• scripts/ramtest.sh ./sei-tendermint/internal/evmonlyapp/... -race green; golangci-lint run (v2.13.2) 0 issues on the package; make fmtcheck clean.

Link to Devin session: https://app.devin.ai/sessions/ff612badcded4aa5914ea408dbb41888
Open in Devin Desktop: https://app.devin.ai/desktop/session/ff612badcded4aa5914ea408dbb41888?variant=devin
Requested by: @bdchatham

[devin-ai-integration]

I'll fix CI failures and address comments from users with write access. I'll skip comments containing "(aside)".

• Disable automatic comment, CI, and merge conflict monitoring

[cursor]

PR Summary

Low Risk
Observability-only refactor with no change to block execution, commit, or ABCI responses.

Overview
Adds OpenTelemetry phase timing around EVM-only FinalizeBlock so wall time outside the executor’s own breakdown is visible in metrics.

A PhaseTimer on evmOnlyApplication emits evmonly_finalize_phase_duration_seconds_total{phase=...} for take_senders, prepare, execute, and tx_results. The finalize path is refactored into finalizeBlockLocked; executeBlockPipelined accepts the timer and marks prepare vs execute. Reset() on exit avoids attributing idle time between blocks.

Measurement only—no consensus or execution behavior change. A new test wires a manual OTel reader and asserts all four phases are recorded after one finalize/commit.

^{Reviewed by Cursor Bugbot for commit c419b39. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

The latest Buf updates on your PR. Results from workflow Buf / buf (pull_request).

Build	Format	Lint	Breaking	Updated (UTC)
✅ passed	✅ passed	✅ passed	✅ passed	Sep 19, 2026, 2:04 PM

[codecov]

Codecov Report

❌ Patch coverage is 94.59459% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 86.17%. Comparing base (7e3ca24) to head (c419b39).
⚠️ Report is 39 commits behind head on giga-1.

Files with missing lines	Patch %	Lines
sei-tendermint/internal/evmonlyapp/app.go	94.59%	2 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           giga-1    #4259       +/-   ##
===========================================
+ Coverage   65.55%   86.17%   +20.61%     
===========================================
  Files        2081        2     -2079     
  Lines      157460      405   -157055     
===========================================
- Hits       103222      349   -102873     
+ Misses      54097       56    -54041     
+ Partials      141        0      -141     

Flag	Coverage Δ
sei-chain	?
sei-chain-pr	86.17% <94.59%> (?)
sei-db	?

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
sei-tendermint/internal/evmonlyapp/app.go	85.51% <94.59%> (+3.74%)	⬆️

... and 2079 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[seidroid]

Measurement-only change that adds a FinalizeBlock phase timer; the instrumentation is correct and the new test genuinely covers all four phases. One minor issue: the deferred Reset runs after the executor mutex is released, so the comment's "only touched with that lock held" invariant does not hold.

Findings: 0 blocking | 1 non-blocking | 1 posted inline

Blockers

• None at the file/PR level.

Non-blocking

• None at the file/PR level.
• 1 suggestion(s)/nit(s) flagged inline on specific lines.