Feature request: Option for multiscale images opened with aszarr to be valid OME-Zarr #335

Right now using aszarr returns zarr that are almost valid OME-Zarr, but not quite.

For example, for a multiscale (pyramidal) image, the multiscales metadata follows ome-zarr. But, the arrays after level 0 have axes names like Y1, X1 (where 1 is the level) that don't follow:

The “dimension_names” attribute MUST be included in the zarr.json of the Zarr array of a multiscale level and MUST match the names in the “axes” metadata.

see: https://ngff.openmicroscopy.org/specifications/0.5/index.html#axes-metadata
(I think the current behavior is for xarray usage, which is totally understandable.)

Also, for the case of an RGB image, the axis order is returned as YXC, which is unfortunately invalid, as C needs to be before spatial axes:

The order of the entries MUST correspond to the order of dimensions of the zarr arrays. In addition, the entries MUST be ordered by “type” where the “time” axis must come first (if present), followed by the “channel” or custom axis (if present) and the axes of type “space”.

See: https://ngff.openmicroscopy.org/specifications/0.5/index.html#multiscales-metadata
The transpose codec does work to resolve this, i've tested it in a wrapper.

All of this can be worked around using a wrapper, but would be really nice if there was a built in option to get valid OME-zarr out-of-the-box.

(an omero block with channel metadata would be nice too, but that's just window dressing)