is 'When to use COG versus Zarr?' Table correct?

[guillaumeeb]

In https://eopf-toolkit.github.io/eopf-101/01_about_eopf/12_about_cloudoptimized_formats.html#when-to-use-cog-versus-zarr, the table indicates features that does not seem correct to me:

• Why is COG access described at sequential? To my knowledge COG can be read in parallel,
• Compression: whole file for COG. COG compression is done chunk by chunk. Do you mean compression codec can't be chosen at a per chunk level?
• Sclae: again, not sure COG is correct, you can have overview and thus different resolutions in the same file, can't you?