Question: why do we need to pass actual schema to the constructor to flink if the schemas are in the schema registry?

[bbaja42]

In https://github.com/awslabs/aws-glue-schema-registry?tab=readme-ov-file#flink-kafka-consumer-with-avro-format
when flink needs to get generic Avro record, why do we need actual schema in

        GlueSchemaRegistryAvroDeserializationSchema.forGeneric(schema, configs),

When message is read from kafka, the magic bytes are used to find the schema ID from schema registry, which is then used to fetch the actual schema.
If the actual schema is in the schema registry, why do we need to still pass the schema when creating Flink source?

In our case, we have large number of schemas, and would like to have a single source that can read all of them.