Deserializer should expose the schema id, as it is needed for schema evolution logic

[wernerdaehn]

When a schema transitions from schema version A to schema version B, often something must be done, e.g. add the additional column in the target database table. No problem if you use the SchemaRegistryClient yourself, then you can keep track of the schema ids consumed and execute some extra code if a new schema id is found for the first time. But if the consumer is configured to use the Confluent SerDe, this information is not exposed.
There are multiple ways of doing that, easiest would be to add that info to the ConsumerRecord.

Not sure if that makes sense for producers as well but for completeness, should be considered.