refactor(eval): address on-demand review feedback

[jariy17]

Follow-up to #1983: makes on-demand evaluation failures easier to classify and removes review-identified noise.

Changes

• Missing runtime telemetry now throws ResourceNotFoundError.
• Failed CloudWatch queries now throw CloudWatchQueryError with query status and ID metadata.
• Malformed telemetry produces at most one warning per command.
• Repeated validation tests now use test.each.
• Redundant and drift-prone comments were removed.

Decisions

1. Fail fast: one session can require multiple Evaluate calls. Continuing after one fails could return incomplete results while appearing successful.
2. Keep results flat: context.spanContext.sessionId supports per-session filtering without changing the SDK response shape.
3. Defer a generic poller: extract one when another production caller needs the same behavior.
4. Keep window resolvers separate: batch feeds the API's DataSourceConfig, including CloudWatch log groups and onlineEvaluationConfigArn; on-demand feeds client-side Logs Insights.

Verification

• Build, typecheck, lint, changed-file formatting, and focused tests pass.
• Full suite: 1,373 tests pass.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.01%. Comparing base (a9d34be) to head (1779d53).
⚠️ Report is 2 commits behind head on refactor.

Additional details and impacted files

@@             Coverage Diff              @@
##           refactor    #2007      +/-   ##
============================================
+ Coverage     96.96%   97.01%   +0.04%     
============================================
  Files           366      366              
  Lines         21179    21191      +12     
============================================
+ Hits          20536    20558      +22     
+ Misses          643      633      -10     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 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.

[Hweinstock]

thanks for coming back to this! one question on the testing pattern, otherwise lgtm.

[Hweinstock]

should this use the fixture pattern? Might be a little harder to trigger these cases, but seems possible unless i'm missing something.

[Hweinstock]

is this from this PR?

Run bun run format:check
$ prettier --check .
Checking formatting...
[warn] src/core/types.tsx
[warn] Code style issues found in the above file. Run Prettier with --write to fix.
error: script "format:check" exited with code 1