to_json segfaults with timezone-aware datetimes

[davidszotten]

>>> import pandas as pd; import pytz; import datetime; faulthandler
>>> faulthandler.enable()
>>> pd.DataFrame.from_records([{'ts': datetime.datetime(2015,1,1, tzinfo=pytz.UTC)}]).to_json()

Fatal Python error: Segmentation fault

Current thread 0x00007fff73afd300 (most recent call first):
  File "<snip>lib/python2.7/site-packages/pandas/io/json.py", line 78 in write
  File "<snip>lib/python2.7/site-packages/pandas/io/json.py", line 35 in to_json
  File "<snip>lib/python2.7/site-packages/pandas/core/generic.py", line 893 in to_json
  File "<stdin>", line 1 in <module>
Segmentation fault: 11

[jreback]

similarly to #10778 it is waiting for an implementation

[kawochen]

should there be a validation somewhere so that it doesn't segfault even with unsupported types?

[jreback]

@kawochen IIRC their is supposed to be something that does exactly that (and use the defautl handler in that case). maybe its not working (or borking on this check)

[jreback]

cc @Komnomnomnom