Separate GRIB indexing and the building of datasets
This MR separates the creation of GRIB indices and building a dataset based in these indices.
In detail this MR:
- adds a function to create a per-file GRIB index in JSONLines format
- adds convenience functions to parse a single/multiple GRIB indices
- revises
grib2kerchunk_refs
to handle the new GRIB index format - extends the usability of
grib2kerchunk_refs
to support 2d and 3d fields in the same dataset