Towards a Python 3 compatible release

[jhamman]

I'm wondering what the timetable is to making a full PyNio release with Python 3 support? At the end of #10, @bladwig1 suggested:

an official release on conda-forge won't take place until a large amount of bug fixes related to compound data types, HDF5, etc are included.

I don't see any open issues related to HDF5 and compound data type bugs that are open right now so its hard to gauge how that progress is coming.

For context, Xarray, which supports PyNio as a backend is dropping Python 2 support at the end of 2018. Ideally, we would have a Python 3 release well before then.

xref: pydata/xarray#1830, conda-forge/pynio-feedstock#23

[bladwig1]

I'll be able to turn my attention back to PyNIO in March. Once #15 is fixed, we'll probably do a release, which will be the same as the previous version but with Python 3.x. The HDF5 issues are recorded in an internal database for NCL (same underlying code), and will most likely require some changes to PyNIO to work, so hopefully by the summer.

[CAM-Gerlach]

Given the packages have been out on conda-forge for what appears like over a year, is Python 3 considered fully supported, and this issue be closed to make that clear to users? If not, is there something in particular blocking this one? Thanks!