Add feature to l2ss-py to support OB_DAAC lat/lon grid instead of nans for SeaDAS tool

[jamesfwood]

• Needed a specific output to support their SeaDAS tool (https://seadas.gsfc.nasa.gov/)
  • Will probably be a special case that requires an input argument to harmony

[sliu008]

working on it in branch feature/simon_pixel_mask some points brought up by obdaac

1. It's great that you kept the lat and lon in place and nan put the data; Could you please also nan out the l2_flags

2. It's great that you made it a nc file with metadata. Can you make the northernmost_lattitude, southernmost_ lattitude, eastmost_longitude and westmost_longitude reflect the actual value?

3. Is it possible that you make the current way of subset default and add an option that you don't nan out the data and L2_flags (don't cut the corners), for SeaDAS does the subset this way

[frankinspace]

1. The l2_flags were not masked because they did not have _Fill_value. @sliu008 will update to use netCDF4.default_fillvals in this case
2. I recall having discussions in the past about whether the services should update global metadata or not; perhaps we should get concurrence on this one just because it would be a change that would effect every collection
3. If I understand correctly, this is asking to add an option to select whether data should be masked or not after cutting. We can add an option to l2ss-py for this but would need to coordinate with harmony team to implement the option for the api