probably temporary, may not really be the best idea

but otherwise get

  cannot pickle '_thread.lock' object

in the joblib parallel setup

normally don't use parallel for this anyway, since there are only
6 time steps and it is pretty fast in serial

since over 100 MB it hits the virus check page though

Also tried this longer command, which supposedly did it in the past

["wget", "--no-verbose", "--no-check-certificate", url, "-O", (HERE / fn).as_posix(), "-r", "-A", "'uc*'", "-e", "robots=off", "-nd"]

but didn't work now

plotting the data and seeing how the contouring could work

they are the same geo area and such, might as well

wget command for the large file that worked before doesn't anymore

there are apparently workarounds that work in 2024
https://medium.com/geekculture/wget-large-files-from-google-drive-336ba2e1c991

but gdown seems the safer way to go

and hide the output of the first (import) cell

so as not to see the

ERROR 1: PROJ: proj_create_from_database ... failed

message

note: had to use :func: or the functions stopped getting linked
with the addition of a :doc:

seems :ref: works fine without this modification, but then get
annoying anchor links that aren't needed

saw this warning in RTD:
tams/data.py:238: FutureWarning: The return type of `Dataset.dims` will be changed to return a set of dimension names in future, in order to be more consistent with `DataArray.dims`. To access a mapping from dimension names to lengths, please use `Dataset.sizes`.

and crossref the example

also str agg wasn't working I think,
but now it should

but really should type it properly
since single callable should be supported as well