refactor(project): simplify code-based evaluator scaffolding

[jariy17]

Summary

• remove --metric and --model from project add evaluator code-based
• always scaffold the generic Python evaluator for managed code-based evaluators
• remove the bundled DeepEval and AutoEvals templates
• describe --timeout-seconds as the evaluator timeout

Compatibility

• existing evaluator project specs are unchanged
• scripts using the removed flags now receive an unknown-option error

Testing

• bun test (3171 passed)
• bun run typecheck
• bun run lint:check
• bun run format:check
• bun run build

Live verification

Verified with a temporary evaluator-only project in an exploratory AWS account in us-west-2:

• scaffolded a managed code-based evaluator with --timeout-seconds 123
• synthesized CloudFormation with Lambda Timeout and evaluator LambdaTimeoutInSeconds both set to 123
• deployed successfully; the evaluator reached ACTIVE status
• confirmed the deployed Lambda was Active with Timeout 123
• invoked the packaged evaluator Lambda with a representative evaluator event and received a Pass response with HTTP 200
• removed the project and confirmed the stack reached DELETE_COMPLETE and both resources no longer exist

[agentcore-devx-automation]

Claude Security Review: no high-confidence findings. (run)

[agentcore-devx-automation]

AgentCore Harness Review

Verdict: Looks good

The change is self-consistent and cleanly scoped to what the description promises:

• --metric and --model flags are fully removed from handlers/project/add/evaluator/code-based/index.ts, along with their parsers (parseMetric, resolveBedrock Model).
• EVALUATOR_LIBRARIES / EvaluatorLibrary are dropped from handlers/project/types.ts, and ManagedEvaluatorScaffoldInput no longer carries metric/model.
• core/project/templates/evaluator.ts collapses to a single evaluators/python-lambda asset dir with a Name-only render context.
• The deepeval-lambda and autoevals-lambda asset trees are removed; no remaining references in src/ (only the test line that asserts the removed flags reject).
• Tests are updated to cover managed-with-timeout, managed-default, --timeout-seconds incompatibility with --lambda-arn, and rejection of the removed flags via commander.unknownOption.
• No TUI screen depended on the removed inputs (isValidBedrockModelId is retained for the llm-as-a-judge handler and TUI, which still need it).
• Existing on-disk evaluator specs are unaffected — metric/model were scaffolder inputs only, not part of the Evaluator schema.

No changes required before merge.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.03%. Comparing base (d9ab959) to head (52787db).
⚠️ Report is 5 commits behind head on refactor.

Additional details and impacted files

@@             Coverage Diff              @@
##           refactor    #2255      +/-   ##
============================================
- Coverage     97.04%   97.03%   -0.01%     
============================================
  Files           566      566              
  Lines         39409    39345      -64     
============================================
- Hits          38244    38180      -64     
  Misses         1165     1165              

☔ 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.

[agentcore-devx-automation]

Claude Security Review: no high-confidence findings. (run)