Updates for v0.5

[gutmann]

• Updates to interpolation code:
  • now uses triangulation and works better on highly skewed input grids
  • option to use Nearest neighbor interpolation is implemented now
• change output files to HDF5/netcdf4
• added erai2gard helper (preliminary, not recommended for use yet)
• fixed several bugs, particularly around n_log_analog and logistic_from_analog options
• improve parallelization
• now accepts proleptic_gregorian for a calendar name
• checks actual number of threads for printing stats at the end
• significant update to post_proc script
• additional documentation

[gutmann]

What are your concerns with the post_proc changes? Would it make sense to document them in an issue instead of this PR?