Filter out timezones not supported by Python (NVIDIA#13289)

Fixes NVIDIA#13285

### Description
The root cause is the inconsistent between Python and JVM: JVM supports
'America/Coyhaique' but Python gets None in the code
`tz.gettz(time_zone)`. And the `TimestampGen` will use
`TimestampNTZType` type which is not expected, then causes error:
`TypeError: can't subtract offset-naive and offset-aware datetimes`

### Fix
Filter out timezones not supported by Python

### Checklists

This PR has:

- [ ] added documentation for new or modified features or behaviors.
- [x] updated the license in the source code files when it is required.
- [x] added new tests or modified existing tests to cover new code
paths.
(Please explain in the PR description how the new code paths are tested,
such as names of the new/existing tests that cover them.)

Please select one of the following options:
- [ ] Performance testing has been performed and its results are added
in the PR description.
- [ ] An issue is filed for performance testing and its link is added in
the PR description. (Select this if performance testing will not be
completed before the PR is submitted.)

---------

Signed-off-by: Chong Gao <[email protected]>
Co-authored-by: Chong Gao <[email protected]>

Author: res-life

integration_tests/src/main/python/timezones.py

@@ -13,6 +13,7 @@
 # limitations under the License.
 
 from conftest import spark_jvm
+from dateutil import tz
 
 fixed_offset_timezones = ["Asia/Shanghai", "UTC", "UTC+0", "UTC-0", "GMT", "GMT+0", "GMT-0", "EST", "MST", "VST"]
 variable_offset_timezones = ["PST", "NST", "AST", "America/Los_Angeles", "America/New_York", "America/Chicago"]
@@ -21,5 +22,6 @@
 
 # Dynamically get supported timezones from JVM.
 # Different JVMs can have different timezones, should not use a constant list here.
-# Note: excludes `America/Coyhaique`, refer to bug: https://github.com/NVIDIA/spark-rapids/issues/13285
-all_timezones = [tz for tz in spark_jvm().java.time.ZoneId.getAvailableZoneIds() if tz != 'America/Coyhaique']
+# Also different Python versions can have different timezones, so we also filter out timezones not supported by Python.
+all_timezones = [zone_name for zone_name in spark_jvm().java.time.ZoneId.getAvailableZoneIds()
+                 if tz.gettz(zone_name) is not None]