I had a brief discussion with mathis.bessa@mongodb.com on serverless about the parseFromScanDef function in abt_lowering. This function extracts a NamespaceStringOrUUID from a ScanDefinition, which sounds more like a getter than a parser.

I think we already construct a NamespaceStringOrUUID much earlier in the request, so we should store that on ScanDefinition directly, instead of converting to a string and back. Classes like NamespaceStringOrUUID and DatabaseName track the tenantId separately from the user-visible database name, which helps avoid ambiguity between a database's internal name and its user-visible name.