Clarify pytz localization steps in documentation

[rhine3]

Current localization requires pytz-localized timestamps to run (but see: #1197). Add a few clarifying notes about this in the documentation:

• Expose the documentation about the need for localized datetimes in the higher level localize_detections function, as currently it's only exposed in create_candidate_events -

  opensoundscape/opensoundscape/localization/synchronized_recorder_array.py

  Lines 328 to 348 in 2f14a07

  	Example of adding `start_timestamp` to multi-index of detections df manually:
  	```python
  	# assume we have a csv with columns file, start_time, end_time, class1, ...
  	# e.g., the output of and opensoundscape CNN prediction workflow
  	detections = pd.read_csv('my_opso_detections.csv',index_col=[0,1,2])
  	# let's assume all files started recording at the same time for this example:
  	import pytz
  	import datetime
  	tz = pytz.timezone('America/New_York')
  	detections = detections.reset_index(drop=False)
  	detections['start_timestamp']=[
  	datetime.datetime(2024,1,1,10,0,0).astimezonoe(tz)+
  	datetime.timedelta(detections.at[i,'start_time'])
  	for i in detections.index
  	]
  	# add the start_timestamp column into the multi-index
  	detections = detections.reset_index(drop=False).set_index(
  	["file", "start_time", "end_time", "start_timestamp"]
  	)
  	```

• See if we can fix the Python code rendering in RTD in this context