Add discard_unknown to schema_registry_encode

Currently, schema registry encode processor does not support encoding messages that have extra/unknown fields. The protobuf processor includes a discard_unknown option to discard fields not present in the protobuf definition.

Desired behavior is for the schema_registry_encode processor to match the protobuf processor behavior and expose a discard_unknown configuration so that the below will not result in an error:

Given a proto definition of:

syntax = "proto3";
package testing;

message Person {
  string first_name = 1;
  string last_name = 2;
}

Today, the following message will fail to schema registry encode:

{
"first_name": "Foo",
"middle_name": "Bar",
"last_name": "Baz"
}

with an error similar to "unmarshal 'Person': proto: (line 1:...): unknown field \"middle_name\"",

Tomorrow, ideally this gets encoded without error, resulting in the protobuf messaged published to Redpanda as:

{
"first_name": "Foo",
"last_name": "Baz"
}

Also, we should be able to chain this together to do something like this:

    - schema_registry_encode:
        url: ${SCHEMA_REGISTRY_URL}
        subject: ${! meta("__kafka_topic") }-value
        refresh_period: 10m
    - catch:
        ...
        - schema_registry_encode:
            url: ${SCHEMA_REGISTRY_URL}
            subject: ${! meta("__kafka_topic_with_evolution") }-value
            refresh_period: 10m
            discard_unknown: true

Start at the schema_registry_encode processor and compare its configuration and behavior with the protobuf processor's discard_unknown option. Add the corresponding configuration behavior, then verify that the Person example encodes while omitting middle_name and that the option works in the shown catch chain.