feat(mip8): storage perf-regression suite + consume timing report

[pdobacz]

I wanted to get the performance regression stack into main, but it resulted huge. I trimmed it down considerably over the course of these commits, but still huge.

I am hesitant whether we need it, but not rejecting the idea, since most of this change is purely additive and would be easy to get rid of.

There's a bunch of tests and also the benchmark tests are going to be filled on CI to ensure they do not rot [that badly].

Happy to hear that it's better not to merge this and re-create when new need arises.

Greptile Summary

The PR adds a MIP-8 storage performance-regression suite and a reporting pipeline that collects Monad runloop block timings and compares MONAD_NINE with MONAD_TEN.

• Adds pageified-storage benchmark fixtures, CI fill configuration, benchmark documentation, and an orchestration script.
• Adds direct-consume timing collection and Monad runloop timing parsing.
• Adds statistical reporting with paired Wilcoxon tests and Benjamini–Hochberg adjustment.
• Updates merged-fixture consumption so unnamed multi-fixture files are streamed and every fixture is verified.

Confidence Score: 5/5

The PR appears safe to merge.

No blocking failure remains; the current unnamed-fixture path iterates and verifies every entry in a merged fixture file, resolving the prior selection and verification-abort reports.

Important Files Changed

Filename	Overview
packages/testing/src/execution_testing/client_clis/clis/monad.py	Streams every fixture from unnamed merged files, verifies each independently, and exposes parsed per-block runloop timings.
packages/testing/src/execution_testing/cli/pytest_commands/plugins/consume/direct/timing_report.py	Adds serial direct-consume timing collection and session-level CSV reporting.
packages/testing/src/execution_testing/cli/perf_regression.py	Builds paired NINE-versus-TEN statistical comparisons and renders the results as Markdown.
tests/benchmark/stateful/mip8_pageified_storage/test_perf_regression.py	Defines the MIP-8 pageified-storage benchmark workloads and fixture sizing behavior.
scripts/perf_cycle.sh	Orchestrates fixture generation, repeated timing runs, and regression-report creation.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart LR
    A[MIP-8 benchmark tests] --> B[Fill NINE and TEN fixtures]
    B --> C[Monad runloop consume passes]
    C --> D[Parse per-block timing logs]
    D --> E[timing_consume.csv per run]
    E --> F[Pair NINE and TEN samples]
    F --> G[Wilcoxon tests and BH adjustment]
    G --> H[Markdown regression report]

Loading

_{Reviews (4): Last reviewed commit: "perf: follow the harness into the repo u..." | Re-trigger Greptile}