Ntv-pandas

--------------------
- extension to multidimensional interfaces
- Xarray interface (`to_xarray` method, `from_xarray` function)
- scipp interface (`to_scipp` method, `from_scipp` function)
- ENH 4 5 7 8 9

--------------------
- bug setup

--------------------
- add `to_analysis` function to use `TAB-analysis`module
- add `check_relation` function to use `TAB-dataset.check_relation` method
- add pandas accessor for `to_analysis` and `check_relation`

--------------------
- move `decode_ntv_tab` method in `ntv_util` module ('json_ntv')

--------------------
- add User Guide
- add `equals` method (check names and cat.categories dtype)
- add pandas accessor for `to_json`, `equals` and `as_def_type`
- BUG 3: categorical with a single value

--------------------
- add Table Schema interface
- Table Schema type / format supported:
- numerical: integer/default, boolean/default, number/default
- json: object/default, array/default, string/default, string/uri, string/email
- datation: datetime/default, date/default, time/default, yearmonth/default, year/default
- location: geopint/array, geojson/default
- additional type / format supported (Table Schema extension):
- numerical: number/floatxx, integer/intxx, integer/uintxx
- json: string/file, object/null, object/object
- datation: date/day, date/wday, date/yday, date/week, time/hour, time/minute, time/second
- location: geojson/geometry, geojson/polygon, geojson/line
- correspondance between `NTVtype` and `type`/ `format` in `ntv_table.ini`