[BUG] Remove pytest x-failures as pandas bug fixes are merged

[brandon-b-miller]

Describe the bug
In the process of upgrading to Pandas 1.0, we were forced to x-fail some tests due to various changes in Pandas behavior. Some of these were intentional changes, and more discussion is warranted around if and to what extent we should attempt to replicate the behavior. Others were simply bugs that we reported, or were already being worked on when we noticed them. This issue tracks these cases, so that we can re-enable the relevant tests as things are fixed in patch versions of pandas.

• test_dataframe.py::test_isin_dataframe, BUG: DataFrame.isin fails when other is a categorical series pandas-dev/pandas#34256
• test_json::test_json_writer, read_json with typ="series" of json list of bools results in timestamps/Exception pandas-dev/pandas#31464
• test_string::test_string_cat, BUG: str.cat produces NaNs when others is an Index pandas-dev/pandas#33425
• test_rolling::test_rolling_series_basic, test_rolling_dataframe_basic, [BUG] Rolling count aggregations produce different results than Pandas 1.0+ #5580

[vyasr]

Closing as a duplicate of #7386.