Filenames without suffix, e.g. in Zarr V2 stores, are rejected

[juntyr]

It is possible to open a Zarr V2 dataset by first translating it to kerchunk with

kerchunk.zarr.ZarrToZarr("ds.zarr").translate()

and then opening it with

virtualizarr.open_virtual_dataset("ds.json", filetype="kerchunk")

However, Zarr uses files without suffix for its data chunks, e.g. var/0, which are rejected during path-to-URI normalization. When I patched out this check the dataset loaded correctly. Would it be possible to disable the check?

[TomNicholas]

Thanks for raising this!

FYI we do have a dedicated Zarr reader coming in #271.

So the question becomes: "is there any other reason to disable this check other than for opening zarr (which will soon be handled by the dedicated reader)?"

cc @norlandrhagen