feat(eda): add clear_cache option to /eda endpoint

[mrmaloof]

Adds support for ?clear-cache=1 on GET /institutions/{inst_id}/batch/{batch_id}/eda. When present, the handler removes the in-memory TTL cache entry for that institution and batch before resolving the response, so the next response is recomputed from storage instead of returning a cached payload.

changes

• Optional query parameter clear-cache (URL query only; path and response shape unchanged).
• If clear-cache equals "1", EDA_CACHE.pop(cache_key, None) runs after the batch is validated and before the normal cache get/set flow.
• Query is declared as Annotated[..., Query(alias="clear-cache")] = None so the default lives only on the parameter, matching FastAPI’s rules for Annotated + Query.

context

EDA results are cached to avoid repeated reads and heavy computation. Operators and developers sometimes need a forced refresh after data or logic changes without waiting for TTL expiry or restarting the service.

---

• To see the specific tasks where the Asana app for GitHub is being used, see below:
  • https://app.asana.com/0/0/1214645011349437