Improve backend coverage for utility serialization and inbox queue behavior

[Copilot]

Coverage analysis showed thin test depth around shared utility serialization logic and message queue/debounce behavior, leaving important edge paths unverified. This PR adds targeted tests for those high-leverage paths and hardens one serialization branch discovered during coverage-driven testing.

• Coverage focus: backend/utils.py

  • Added direct tests for:
    • UTC timestamp normalization/parsing (ts_fmt, parse_db_iso_utc)
    • timezone resolution + invalid timezone fallback (app_timezone, now_*)
    • DataFrame JSON sanitization (dataframe_to_json_safe) across NaN, Inf, datetime, and empty-frame inputs
    • recursive dict/list scalar serialization (clean_dict_for_json, serialize_value)

• Coverage focus: backend/messaging/inbox.py

  • Added queue behavior tests for:
    • push/pop roundtrip and empty drains
    • bounded queue overflow semantics (drop oldest)
    • debounce merge behavior within window
    • no merge across sender/channel boundaries
    • wait_and_pop_all burst-drain behavior

• Targeted runtime hardening

  • Updated serialize_value to handle np.ndarray before pd.isna(...) checks, avoiding ambiguous truth-value failures on array inputs.
  • Simplified control flow to early-return style for readability.

def serialize_value(value: Any) -> Any:
    if isinstance(value, np.ndarray):
        return value.tolist()
    if isinstance(value, (pd.Timestamp, np.datetime64)):
        return ts_fmt(value.to_pydatetime()) if hasattr(value, 'to_pydatetime') else str(value)
    if isinstance(value, (np.integer, np.floating)):
        if np.isnan(value) or np.isinf(value):
            return None
        return value.item()
    try:
        if pd.isna(value):
            return None
    except (TypeError, ValueError):
        pass
    return value

[Copilot]

Pull request overview

This PR improves backend test coverage around shared utility serialization/timestamp helpers and async inbox queue behavior, and hardens serialize_value to avoid ambiguous truth-value errors on NumPy arrays.

Changes:

• Added unit tests for UTC timestamp formatting/parsing, timezone selection/fallback, DataFrame-to-JSON sanitization, and recursive scalar serialization.
• Added async tests for InboxQueue push/pop, bounded overflow (drop-oldest), debounce merge rules, and wait_and_pop_all drain behavior.
• Updated serialize_value to handle np.ndarray before pd.isna(...) and to use an early-return flow with guarded pd.isna evaluation.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

File	Description
tests/test_utils.py	Adds coverage for timestamp/timezone helpers and JSON-safe serialization utilities.
tests/test_inbox_queue.py	Adds coverage for inbox queue draining, maxsize overflow behavior, and debounce semantics.
backend/utils.py	Hardens serialize_value for NumPy arrays and simplifies the control flow.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

These assertions depend on IANA tzdata being available for ZoneInfo("Asia/Shanghai"). The repo doesn’t declare a tzdata dependency and the Docker image is based on python:*-slim (often missing /usr/share/zoneinfo), so this test can fail with ZoneInfoNotFoundError in some environments. Consider using a fixed-offset datetime.timezone(timedelta(hours=8)) here, or skipping the test when ZoneInfo("Asia/Shanghai") isn’t available.

[Copilot]

This test assumes ZoneInfo("Asia/Shanghai") is resolvable on the runner. On environments without tzdata (common for slim Docker bases), utils.app_timezone() can’t reliably return non-UTC zones and this test will fail. Consider guarding with a try/except ZoneInfoNotFoundError and pytest.skip(...), or ensuring tzdata is installed as part of the test/runtime deps.