Consider recommending against the use of add_offset and scale_factor

[asinghvi17]

Assuming that we want to be as close to Zarr as possible here, a plain Zarr reader would ideally provide values as close as possible to the true value (barring missings etc) for a GeoZarr file.

As such, since Zarr already has a scale-offset filter, would you consider recommending using that in place of the CF attributes in the README?

[christophenoel]

Thank you for that brilliant question! I still hope that we will complete a first version of the specification by 2025. Given the current pace, I would prefer to address such questions in a second version.

However, I do believe the specification would likely recommend encoding optimisation (such as scale and offset) through filters, provided Zarr officially releases them.