Pycurious

Fixes a number of parallel bugs and introduces a new functions in the `mapping` module:

- `ungrid` for interpolating to unstructured points
- `export_netcdf4` for exporting a dataset to netcdf4
- `import_netcdf4` for importing a dataset to netcdf4

Ben Mather and Robert Delhaye

Formal release after [JOSS review](https://github.com/openjournals/joss-reviews/issues/1544).

This version of `pycurious` implements the following functionality:

- Anaconda installation support (and fixes for Cython build issues)
- Checking if windows that exceed the extent of the domain
- Correct propagation of errors with Tanaka *et al.* 1999 approach
- Refactor code to compute the radial power spectrum
- Updated Docstrings and Jupyter notebooks
- Updated test suite (using `pytest`) to test for expected functionality of core routines
- PEP8 compliant code

Depreciated functions:

- `CurieGrid.radial_spectrum_log`; instead use `CurieGrid.radial_spectrum` and pass the argument `power=0.5` for the same functionality.

Release to coincide with JOSS submission

Implemented

- `pycurious.download` module for downloading external sources
- Automated testing with `pytest`
- Automated docs generation with `pdoc`
- Minor bug fixes in several places

Still to do before JOSS submission:

- Neater documentation
- Iron out any deployment issues

- Python 3 support
- Parallel routines based on Python's multiprocessing module
- Clearer implementation of Bayesian analysis tools