feat: add TaskResultStore for caching and replaying task execution results

[afarntrog]

Description

Add TaskResultStore protocol for caching task execution results across experiment runs. This enables skipping expensive task re-execution (e.g., LLM calls) when iterating on evaluators, by loading cached EvaluationData from a pluggable storage backend. In addition, refactor the multi hundred line _worker method.

• Introduce TaskResultStore protocol with load/save methods that implementations can back with any storage (files, S3, databases, etc.)
• Accept an optional task_result_store parameter in run_evaluations and run_evaluations_async
• When a store is provided, load cached results before executing the task and save results after execution
• Validate that all cases have unique, non-None names when a store is provided
• Refactor _worker by extracting _execute_task and _run_evaluator into standalone methods for readability
• Export TaskResultStore from the top-level strands_evals package

Related Issues

#94

Documentation PR

Type of Change

New feature

Testing

Added 5 unit tests in tests/strands_evals/test_experiment.py under a new TestTaskResultStore class covering:

• Task execution and result saving on first run with an empty store
• Cached result loading on subsequent runs (task not called)
• Validation that cases must have names when a store is provided
• Validation that case names must be unique when a store is provided
• Default behavior without a store remains unchanged

Uses a DictTaskResultStore in-memory implementation for test isolation.

• I ran hatch run prepare

Checklist

• I have read the CONTRIBUTING document
• I have added any necessary tests that prove my fix is effective or my feature works
• I have updated the documentation accordingly
• I have added an appropriate example to the documentation to outline the feature, or no new docs are needed
• My changes generate no new warnings
• Any dependent changes have been merged and published

---

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[poshinchen]

Is the next step implementing the local store (save and load json)?

[afarntrog]

Is the next step implementing the local store (save and load json)?

Yes