Fix/epsg3857

[Karinon]

This hopefully fixes #176.

The following datasets didn't work before and should now work correctly:

https://fix-epsg3857.gridlook.pages.dev/#https://atlantis-vis-o.s3-ext.jc.rl.ac.uk/nemotest101/pyramid2/T1d/sos_abs.zarr

https://fix-epsg3857.gridlook.pages.dev/#https://carbonplan-maps.s3.us-west-2.amazonaws.com/v2/demo/4d/tavg-prec-month

Also extended support for some curvilinear datasets:

https://fix-epsg3857.gridlook.pages.dev/#https://carbonplan-share.s3.us-west-2.amazonaws.com/zarr-layer-examples/antarctic_era5.zarr

https://fix-epsg3857.gridlook.pages.dev/#https://whiffle-wins50-data.s3.amazonaws.com/data/whiffle_wins50_2020_scenario_fields.zarr @eeholmes this one didn't work on your fork either

CAUTION: A lot of data
https://fix-epsg3857.gridlook.pages.dev/#https://usgs.osn.mghpcc.org/hytest/conus404-biasadjusted/conus404-biasadjusted_daily.zarr

The following reference datasets were expected to still run correctly:

https://fix-epsg3857.gridlook.pages.dev/#https://object-store.os-api.cci2.ecmwf.int/mlcast-source-datasets/radklim/v0.1.0/hourly.zarr/

https://fix-epsg3857.gridlook.pages.dev/#https://storage.googleapis.com/cmip6/CMIP6/CMIP/MOHC/HadGEM3-GC31-LL/historical/r1i1p1f3/Omon/uo/gn/v20190624/

https://fix-epsg3857.gridlook.pages.dev/#https://storage.googleapis.com/cmip6/CMIP6/CMIP/NOAA-GFDL/GFDL-CM4/1pctCO2/r1i1p1f1/Omon/tos/gn/v20180701/

https://fix-epsg3857.gridlook.pages.dev/#https://data.dynamical.org/noaa/hrrr/forecast-48-hour/latest.zarr

https://fix-epsg3857.gridlook.pages.dev/#https://atlantis-vis-o.s3-ext.jc.rl.ac.uk/noc-npd-era5-demo/npd-eorca025-era5v1/gn/T1y_4d/thetao_con

[cloudflare-workers-and-pages]

Deploying gridlook with    Cloudflare Pages

Latest commit:	68ff001
Status:	✅  Deploy successful!
Preview URL:	https://83b2e49b.gridlook.pages.dev
Branch Preview URL:	https://fix-epsg3857.gridlook.pages.dev

View logs

[wachsylon]

@d70-t Do we want to enable rendering these datasets?

Gridlook mostly supports CF and we could go further and support Geozarr. In this case, I suppose we want datasets to look like this. What I can say is

• grid_mapping parameters are not available in the datasets (not CF compliant)
• proj: parameters are not available in the datasets and the code also does not use the specs to identify the projections, right? (not Geozarr compliant)

Imo, it would still be OK to merge if we can agree on any convention policy / priority order, e.g.:

1. CF
2. Geozarr
3. Whatever Gridlook wants to do

But I think we need some changes to first look for proj: things before just checking for x and y, right?