Page LoCoMo ingest wait across memory list results

[Jacobzqy]

Summary

• page through session-scoped memory list results while waiting for LoCoMo
  ingest writes
• count all matching memories instead of assuming a single list request can
  observe every write

Why

The LoCoMo harness waits for all raw memory writes for a sample to become
searchable
before evaluation continues. The server caps /memories list results, but the
wait
logic previously requested limit=expectedCount and compared only that single
page.

For larger samples like conv-26, the expected write count exceeds the API
page size,
so the harness would time out even after writes succeeded.

Verification

• cd benchmark/locomo && npm run typecheck
• reproduced the timeout on conv-26 before the fix
• verified the new wait logic pages through results instead of stalling at the
  API cap

[netlify]

👷 Deploy request for mem9 pending review.

A Netlify team Owner will need to approve the deploy before you can run your build.

Are you a team Owner? Visit the deploys page to approve it →

Need more help? Learn more in the Netlify docs →

Name	Link
🔨 Latest commit	7552437