Fast pfb

[wh3248]

Add a new file "fast_pfb_reader.py" that implements an performance optimized function to read pfb files.
The function is called read_files() and accepts the same arguments as parflow function read_pfb_sequence().
It is optimized to minimize the number of IO reads by computing subgrid offsets in functions rather than file lookups.
It also maximizes the number of threads used to read parallel pfb files.

The new function is integrated into hf_hydrodata.gridded.py to replace the old call to read_pfb_sequences.
There are unit tests that run only on delaware/verde with /hydrodata, but the unit tests are skipped it run on remote machines.

This also contains some changes to allow the hf_hydrodata util unit tests do not fail when run remotely.

[amy-defnet]

Thanks, Bill! I'm not as close to the details on the subgrid reading, but a high-level review of that looks good. The new tests are nice to ensure we're still returning the same answer as the read_pfb_sequence implementation.