fix: Make TaskEvaluation.quality optional with default value

[Mukhsin0508]

Problem

TaskEvaluation Pydantic validation fails when LLM streaming responses omit the required quality field, causing memory save failures. This matches the issue described in #3915.

Solution

• Made quality field optional with default value of 5.0
• Improved evaluation prompt to emphasize quality requirement
• Added comprehensive tests

Testing

• Added tests for missing quality field (defaults to 5.0)
• Added tests for provided quality field (backward compatible)
• Verified no linting errors

Impact

• Fixes validation errors preventing memory saves
• Backward compatible (existing code continues to work)
• Low risk (LongTermMemoryItem already accepts None for quality)

Fixes #3915

---

[!NOTE] Set a 5.0 default for TaskEvaluation.quality, tighten the evaluation prompt requirements, and add tests covering missing/provided quality and JSON parsing.

• Utilities
  • TaskEvaluation: set quality default to 5.0 and update its description.
• Evaluator Prompt
  • TaskEvaluator.evaluate: require ALL fields; emphasize numeric, required quality score in instructions.
• Tests
  • Add tests verifying default quality when omitted, handling when provided, and model_validate_json with partial JSON; update imports accordingly.

^{Written by Cursor Bugbot for commit 5d1f44449be3e3dd528fbb89f3f7845ec4c94e93. This will update automatically on new commits. Configure here.}