At the moment the field ids are encoded as string, which takes up more storage space than is required, and may expose us to fetcher bugs in the future similar to #2810 (I'm pretty confident we are safe atm).

It will likely be necessary to remove the current "C" field ID before doing so, as the binary encoded uint32 values will clash with "C" and otherwise make encoding/decoding harder than it needs to be. "C" isn't even really used anymore, its just the DataStoreKey is misused as HeadStoreKey in someplaces, which does still use "C".