Composite Store unit tests

[cody-littley]

Describe your changes and provide context

New suite of unit tests for composite.Store

[cursor]

PR Summary

Low Risk
Test-only additions with no production code changes; increases CI time for the composite package but does not affect runtime behavior.

Overview
Adds a deterministic, oracle-based randomized test suite for CompositeCommitStore (~2k lines of test-only code in random_test_framework_test.go and random_test.go).

The harness drives the full store API (commit, read, iterate, proofs, rollback, export/import, historical LoadVersion) against an in-memory reference model, with deep checks on memiavl vs flatkv placement, physical flatkv rows (including EVM account merge), migration metadata, commit info, and ICS-23 proof routing. Workloads cover bank/staking/evm stores, intra-block last-write-wins conflicts, and flatkv-specific key/value shapes.

Eight scenario tests (go test -run Random): five steady-state WriteModes (CRUD, rollback+restart, WAL replay restart, read-only historical snapshot, state-sync clone, dual-backend crash reconcile) and three migration paths (EVM, all-but-bank, bank) including mid-migration rollback, crash reconcile, and clone-while-in-flight.

^{Reviewed by Cursor Bugbot for commit 0aeffcb. 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	Jun 11, 2026, 6:41 PM

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 58.38%. Comparing base (9833853) to head (0aeffcb).
⚠️ Report is 6 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3576      +/-   ##
==========================================
- Coverage   59.18%   58.38%   -0.81%     
==========================================
  Files        2210     2139      -71     
  Lines      183250   175208    -8042     
==========================================
- Hits       108460   102292    -6168     
+ Misses      64998    63780    -1218     
+ Partials     9792     9136     -656     

Flag	Coverage Δ
sei-db	70.41% <ø> (ø)
sei-db-state-db	?
sei-db-state-db-pr	73.46% <ø> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.
see 72 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.

[blindchaser]

Go map iteration order is randomized per run, so the resulting keys slice order is nondeterministic. Every later keysInUse.Sample(...) indexes into that slice, meaning which keys get read/updated/deleted/sampled depends on map order, not the RNG seed. This contradicts the harness's explicit contract ("a given seed yields a byte-identical apply/commit sequence", "fully reproducible from r's seed", framework header lines 158–160 and 439–440). The guarantee holds for Phase 1 only; it's violated everywhere liveKeySetFromOracle is used — steady-state Phases 2+ (random_test.go:82,124,157) and every migration resume (:368,392,410).

Practical impact: when CI fails and prints Random seed: N, re-running with seed N will not reproduce the failing sequence past the first rollback — the printed seed is misleading exactly when you need it. Fix is small: collect the pairs into a slice, sort by (store, key), then Add in sorted order.

[cody-littley]

change made

[blindchaser]

flakiness: assertMigrationInFlight is seed-dependent and can spuriously fail

[cody-littley]

fixed