DRAFT -- updates for recent changes to hydrofabric, etc.

We made some temporary fixes in order to make the notebooks produce ngen input after some updates to ngen, the input datasets, and other dependencies. (These notebooks are amazing -- thank you @igarousi! and @Castronova!) These are things others are also working on, so these may quickly be obsolete notes. Commits and possibly a PR are forthcoming.

## For the ngen-hydrofabric-subset:
### As temporary fixes:
Bypassed the map by building a simple dataframe from one of the divide objects.
Updated the URL of the geopackage to the lynker-spatial s3 location
In subset.py, build the parquet name for the model_attributes from the new path in Lynker-spatial.com

### Need to do
Adjust ngen.yaml template
Rebuild to use hf_subset altogether

## For the ngen-create-cfe-forcings
### As temporary fix:
Shunt away from Thredds and instead use direct connection to AORC zarr.
Need to figure out LQFRAC

### Need to do
Transition to TEEHR or Ngen-Datastream method of forcing subsetting for better performance

Start by reviewing the ngen-hydrofabric-subset and ngen-create-cfe-forcings notebooks, then inspect subset.py and the ngen.yaml template. Trace the temporary dataset and forcing-source changes described in the issue and identify the current upstream methods before making updates. Done should mean both notebook workflows use the current dependencies without the listed temporary workarounds.

Written by the indexing model from the issue text.