Probe utcoffset() in convert_to_timestamp to catch deferred toordinal

The previous fix only handled NotImplementedError raised during
pd.Timestamp construction, but pandas 3.0 also defers the toordinal
crash to utcoffset() — which pd.DataFrame() calls when aligning a
mixed-tz column. test_query_data exercises this path via
result_set_to_pandas() → pd.DataFrame(...) and still failed.

Probe utcoffset() right after construction; if it raises
NotImplementedError, fall back to a naive Timestamp so the column
becomes datetime64[ms] (no tz) instead of datetime64[ms, <tz>]. NaT
short-circuits the probe (it doesn't support utcoffset()).

Verified: 0 failures across {ms,us,ns} × {UTC,Etc/UTC,Asia/Shanghai,
America/Los_Angeles} × Long.MIN/MAX/0/±1/Int32 boundaries on Python
3.13 and 3.14 with pandas 3.0.3, for both isoformat() and
pd.DataFrame() construction.

Author: JackieTien97

iotdb-client/client-py/iotdb/utils/rpc_utils.py

@@ -79,23 +79,34 @@ def verify_success_with_redirection_for_multi_devices(status: TSStatus, devices:
 
 def convert_to_timestamp(time: int, precision: str, timezone: str):
     try:
-        return pd.Timestamp(time, unit=precision, tz=timezone)
+        ts = pd.Timestamp(time, unit=precision, tz=timezone)
     except OutOfBoundsDatetime:
         try:
-            return pd.Timestamp(time, unit=precision).tz_localize(timezone)
+            ts = pd.Timestamp(time, unit=precision).tz_localize(timezone)
         except NotImplementedError:
             return pd.Timestamp(time, unit=precision)
     except ValueError:
         logger.warning(
             f"Timezone string '{timezone}' cannot be recognized by pandas. "
             f"Falling back to local timezone: '{get_localzone_name()}'."
         )
-        return pd.Timestamp(time, unit=precision, tz=get_localzone_name())
+        ts = pd.Timestamp(time, unit=precision, tz=get_localzone_name())
     except NotImplementedError:
         # Timestamp falls outside Python's stdlib datetime range (year < 1 or
         # > 9999). pandas >= 3.0 cannot attach a timezone to such Timestamps
         # because tz conversion relies on toordinal(). Return naive instead.
         return pd.Timestamp(time, unit=precision)
+    # Construction succeeded but utcoffset() may still fail downstream (e.g.
+    # in pd.DataFrame's tz alignment) when the Timestamp's year is outside
+    # 1..9999 and tz is not the canonical UTC. Probe early and fall back to
+    # naive so callers like result_set_to_pandas() don't blow up. NaT does
+    # not need this check (and does not support utcoffset()).
+    if ts is not pd.NaT:
+        try:
+            ts.utcoffset()
+        except NotImplementedError:
+            return pd.Timestamp(time, unit=precision)
+    return ts
 
 
 unit_map = {