fix(cli): return 404 when adding a session to a missing eval set

[ankit2235]

Link to Issue or Description of Change

2. Or, if no issue exists, describe the change:

Problem:
POST /dev/apps/{app_name}/eval-sets/{eval_set_id}/add-session returns a 500 when the eval set doesn't exist. The endpoint only catches ValueError around add_eval_case, but every eval sets manager (local, GCS, in-memory) raises NotFoundError for a missing eval set, as documented on EvalSetsManager.add_eval_case. NotFoundError isn't a ValueError, so it goes unhandled.

This is already tracked by test_add_session_to_eval_set_unknown_eval_set_is_a_client_error, which was added as a strict xfail in 4912a08 with the reason "add-session maps ValueError, but the managers raise NotFoundError".

Solution:
Catch NotFoundError and return a 404, the same way the update and delete eval case endpoints already do. The existing ValueError -> 400 handling for a duplicate eval case is unchanged. The xfail marker is removed since the test now passes (it's strict, so leaving it would fail the run).

Testing Plan

Unit Tests:

• I have added or updated unit tests for my change.
• All unit tests pass locally.

With the xfail removed and without the fix, the test fails with:

NotFoundError: EvalSet missing_eval_set not found for app test_app.

With the fix:

pytest tests/unittests/cli/test_fast_api.py::test_add_session_to_eval_set_unknown_eval_set_is_a_client_error
1 passed

pytest tests/unittests/cli -n 4
1442 passed, 5 skipped, 2 xfailed

Manual End-to-End (E2E) Tests:

Not run against a live adk web server. The regression test drives the real FastAPI app through TestClient, which covers the endpoint's error mapping.

Checklist

• I have read the CONTRIBUTING.md document.
• I have performed a self-review of my own code.
• I have commented my code, particularly in hard-to-understand areas.
• I have added tests that prove my fix is effective or that my feature works.
• New and existing unit tests pass locally with my changes.
• I have manually tested my changes end-to-end.
• Any dependent changes have been merged and published in downstream modules.

[yang0228]

Reviewed 579d501eed34823e99b6c0c6af415421639cc388. No blocking correctness findings. Catching NotFoundError matches the local, GCS and in-memory managers' contracts and the adjacent update/delete handlers; the existing duplicate-case ValueError handling remains intact.

Independent validation on Python 3.12.13:

• Parent commit: the existing regression fails with uncaught NotFoundError when run with --runxfail; PR head: both add-session tests pass.
• Four additional TestClient probes pass: canonical and legacy routes return exactly 404 for a missing eval set, while successful insertion returns 200 and duplicate insertion remains 400.
• CLI + eval-set-manager suites: 1,548 passed, 7 skipped, 2 xfailed, 5 setup errors. All five errors also reproduce on the parent commit (agentplatform.frameworks is unavailable in my environment), so this is not an all-green suite claim.
• Pre-commit passes for both changed files, with no changes to the checkout.

I did not run the full Python-version matrix or a live adk web server. One non-blocking test-strengthening suggestion is attached.

[yang0228]

Re-reviewed ce679d03762df5315ad44c0470d26029cd829a44 following your update. The regression now asserts exactly 404, addressing my earlier suggestion. The production fix is unchanged from my previous review; no blocking findings.

Fresh validation on Python 3.12.13:

• Both add-session tests pass (2 passed).
• Four additional TestClient contract probes pass: canonical and legacy routes return exactly 404 for a missing eval set; successful insertion remains 200 and duplicate insertion remains 400.
• Pre-commit passes for both changed files, with no changes made by the hooks.

This is a focused follow-up review. I did not rerun the full suite, the Python-version matrix, or a live adk web server.