Migrate all exposed / exported zarr3 configs to use chunk encoding version 2

[MichaelBuessemeyer]

Detailed Description

Currently, the default chuck encoding is used: https://zarr-specs.readthedocs.io/en/latest/v3/core/v3.0.html#chunk-key-encoding
But migrating from default to v2 would enable us to leave out the c. pretending all zarr3 requests regarding the datasets / datalayers for whose we expose zarr.json configs.

Features where zarr.jsons are written are:

• zarr streaming
• zarr export
• maybe more

To implement this, simply change

 "chunk_key_encoding": {
    "name": "default",
    "configuration": {
      "separator": "."
    }
  },

to

"chunk_key_encoding": {
    "name": "v2",
    "configuration": {
      "separator": "."
    }
  },

[fm3]

Zarr streaming and the volume annotation export is (currently) the complete list. Take care that the exported volume annotations have to also be understood correctly by the reupload (both old and new files).

[fm3]

Looks like in #7995 the zarr streaming was already migrated to v2. This accidentally also chagned the zarr.json used when exporting volume annotations.

This means that volume annotations zarr.jsons are currently inconsistent (or rather, don’t accurately describe the binary data).

Because this is a bug, the priority is higher now. I’ll open a PR today changing the volume export to use v2 with no c. prefix. I’ll also adapt the volume zip uploading code to accept chunks with and without c..