fix(tests): isolate mock session payloads in Vertex AI session service tests

[Jinzhengxu]

Please ensure you have read the contribution guide before creating a pull request.

Link to Issue or Description of Change

1. Link to an existing issue (if applicable):

• Closes: N/A (no existing issue; test-only fix, described below following the bug template)

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

Problem:

tests/unittests/sessions/test_vertex_ai_session_service.py is order-dependent.
The mock_api_client_instance fixture registers the module-level
MOCK_SESSION_JSON_1/2/3/PAGE1/PAGE2 dicts directly in session_dict, while
MockAsyncClient._append_event rewrites session_dict[id]['update_time'] in
place. Every test_append_event* test therefore mutates the shared
MOCK_SESSION_JSON_1 (its update_time becomes the appended event's
timestamp), and any later test that compares get_session() against
MOCK_SESSION (built once at import time from the original timestamp) fails:

tests/unittests/sessions/test_vertex_ai_session_service.py:716: in test_get_and_delete_session
    assert (
E   AssertionError: assert Session(id='1...=1734005534.0) == Session(id='1...005532.123456)

In the default file order the append tests happen to run after the comparisons,
so this is invisible in CI today, but it surfaces as soon as the order changes:
with pytest-randomly, the file alone fails under every seed tried (1–6, 1–2
failures each), and pytest tests/unittests -n 8 fails
test_get_and_delete_session and test_delete_session_rejects_other_users_session
under seeds 1 and 2. Running each of the nine test_append_event* tests
immediately before test_get_and_delete_session reproduces the failure
deterministically in a single process.

Solution:

Deep-copy the session payloads in the fixture, exactly as the fixture already
does for the event payloads (copy.deepcopy(MOCK_EVENT_JSON)), so each test
gets its own session_dict entries and in-place updates cannot leak across
tests. No production code is touched.

Testing Plan

Unit Tests:

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

Test-only change; the fixture itself is the fix. Results on this branch:

• pytest tests/unittests/sessions/test_vertex_ai_session_service.py: 54 passed
  in default order and under pytest-randomly seeds 1–6 (previously 1–2
  failures per seed).
• pytest tests/unittests -n 8 with pytest-randomly seeds 1 and 2: 2 of 2 runs green (15099 passed, 84 skipped, 27 xfailed, 2 xpassed each)
  (on main both seeds failed the two tests above).
• pyink --check and isort --check-only on the file: unchanged.

Manual End-to-End (E2E) Tests:

Not applicable (test-only change).

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. (N/A: test-only)
• Any dependent changes have been merged and published in downstream modules. (N/A)

Additional context

Found while running the unit suite under random ordering. The nine polluting
tests are test_append_event, test_append_event_with_compaction,
test_append_event_with_compaction_and_custom_metadata,
test_append_event_with_usage_metadata,
test_append_event_with_usage_metadata_and_compaction,
test_append_event_with_part_metadata_round_trips,
test_append_event_strips_unsupported_part_metadata,
test_append_event_retries_once_on_429 and
test_append_event_fallback_for_older_sdk; each appends an event to session
'1'.