create intake catalog to abstract data locations

[stevehadd]

Currently we are hard coding the data file locations. It would be nice to create a basic intake catalog which can be used to access the data. Different catalogs with the correct paths can then be created for different processing platforms (primarily AWS, azure/pangeo and MO spice).

[stevehadd]

After chatting to Theo:

• We should be able to simply create a YML file for exposing the CSV files https://intake.readthedocs.io/en/latest/quickstart.html
• we may need something more complicated for the NetCDF data, involving a custom intake driver
• we will need one catalog per data location - initially Met Office and Azure, potentially AWS as well
• the location of the yaml file(s) could be either in the XBT repository or next to the data, or both.
• we will need to find out the licensing status of the WMO data before we publish it
• we probably want entries in the catalog for the following, each one a nested intake category
  • raw WOD data (current)
  • preprocessed current WOD data
  • paper data - raw netcdf
  • paper data - preprocessed

[stevehadd]

XBT data is already available through a public AWS bucket from NOA: https://noaa-wod-pds.s3.amazonaws.com/index.html

[stevehadd]

More useful intake information: https://intake.readthedocs.io/en/latest/catalog.html

[stevehadd]

We should probably include a way to generate the catalog files from a template, to accomodate future changes.