Schema Registry: getSchemaById method missing

Hi all,

We are currently migrating from using `@kafkajs/confluent-schema-registry` to `@confluentinc/schemaregistry`, but one element of the deserialization process is causing us an issue with migrating.

For `@kafkajs/confluent-schema-registry`, the deserialization method only requires the buffer to be passed in. Under the hood, it decodes the `registryId` from the payload, and uses the `getSchemaById` method to complete the deserialization of the payload. You can see the code pointer [here](https://github.com/kafkajs/confluent-schema-registry/blob/master/src/SchemaRegistry.ts#L293-L331).

In `@confluentinc/schemaregistry`, deserialization requires both the topic and the buffer to be passed in. It seems like this is because, rather than `getSchemaById`, this library uses `getBySubjectAndId`. The code pointer for that is [here](https://github.com/confluentinc/confluent-kafka-javascript/blob/master/schemaregistry/serde/serde.ts#L340-L351).

Unfortunately in our specific use case, it's going to be a challenge to thread through the topic to the point where we do our message deserialization - would it be possible to change this interface so that it only requires the buffer instead? Or is there a reason we are missing as to why `getBySubjectAndId` is required in this case?

Thanks in advance!

Start by reading schemaregistry/serde/serde.ts around the referenced deserialization code and compare it with the linked SchemaRegistry.ts implementation. Determine whether schema lookup can be performed from the buffer without a topic, then verify that the deserialization interface and its behavior meet that requirement.

Written by the indexing model from the issue text.