NCZarr Support Part I: Local Datasets

[openSourcerer9000]

In response to #672, I've added in the logic to handle Zarr datasets by passing them through to netCDF-C's NCZarr protocol. protocols/zarr.py takes a Zarr dataset specified in any of the following formats and returns a valid NCZarr URI (specified here), to be recognized by netCDF-C:

"http://s3.amazonaws.com/bucket/dataset.zarr"
"http://s3.amazonaws.com/bucket/dataset.zarr"#mode=nczarr,s3
"/home/path/to/dataset.zarr"
Path('/home/path/to/dataset.zarr')
"file:///home/path/to/dataset.zarr"
"file:///home/path/to/dataset.randomExt#mode=nczarr,file"
"file:///home/path/to/dataset.zarr#mode=nczarr,zip"

Note that so far, this will only work on LOCAL datasets, with the default libnetcdf build installed when compliance-checker is set up with conda. NCZarr is also only fully supported in Linux at the moment, I added an OS check to pass this caveat through to a user trying to run on a Zarr from another OS.

Getting S3 support down for netCDF-C is an ongoing effort. Once it's solid, the S3 test that is currently commented out in test_cli.py should pass and it should work on S3 Zarr datasets.

Update: it looks like this is on the home stretch, AWSome!

While I was in test_protocols.py, I also refactored it to use Pytest, continuing the upgrade to Pytest.

#test_protocols.py

id_url = {
# Check that urls with Content-Type header of "application/x-netcdf" can
# successfully be read into memory for checks.
'netcdf_content_type':"https://gliders.ioos.us/erddap/tabledap/amelia-20180501T0000.ncCF?&time%3E=max(time)-1%20hour",
# Tests that a connection can be made to ERDDAP's GridDAP
'erddap':"http://coastwatch.pfeg.noaa.gov/erddap/griddap/osuChlaAnom",
# Tests that a connection can be made to Hyrax
'hyrax':"http://ingria.coas.oregonstate.edu/opendap/hyrax/aggregated/ocean_time_aggregation.ncml",
# Tests that a connection can be made to a remote THREDDS endpoint
'thredds':"http://thredds.ucar.edu/thredds/dodsC/grib/NCEP/GFS/Global_0p25deg_ana/TP",
# Tests that a connection can be made to an SOS endpoint
'sos':"https://data.oceansmap.com/thredds/sos/caricoos_ag/VIA/VIA.ncml",
}

When run they look like:

test_protocols.py::TestProtocols::test_connection[netcdf_content_type] PASSED
test_protocols.py::TestProtocols::test_connection[erddap] PASSED
test_protocols.py::TestProtocols::test_connection[hyrax] PASSED
test_protocols.py::TestProtocols::test_connection[thredds] PASSED
test_protocols.py::TestProtocols::test_connection[sos] PASSED

[ocefpaf]

Closing in favor of #1071.