[Bug]: Beam: exception reading Kafka Avro with Schema registry

### Apache Hop version?

2.4

### Java version?

11

### Operating system

Windows

### What happened?

Migration: https://issues.apache.org/jira/browse/HOP-4111
https://issues.apache.org/jira/browse/HOP-4112

**Consumer**

I was testing HOP pipelines on a local environment with Kubernetes. I generated a HOP fat-jar and deployed it on our flink-kubernetes-operator. As a schema registry we are using apicurio-registry-sql (https://github.com/Apicurio/apicurio-registry) that is fully compatible with Confluent schema registry API.

There was an exception in the task manager while using Beam Kafka Consume for reading Avro messages from Kafka using a schema registry.
Please see the attached example-kafka-avro-consume.hpl. I was running it with the real ip of apicurio-registry and verified that the URL worked.
The exception that I got in the task manager:
```
Caused by: java.lang.ClassNotFoundException: io.confluent.kafka.serializers.AbstractKafkaSchemaSerDe
at java.base/java.net.URLClassLoader.findClass(Unknown Source)
at java.base/java.lang.ClassLoader.loadClass(Unknown Source)
at org.apache.flink.util.FlinkUserCodeClassLoader.loadClassWithoutExceptionHandling(FlinkUserCodeClassLoader.java:64)
at org.apache.flink.util.ChildFirstClassLoader.loadClassWithoutExceptionHandling(ChildFirstClassLoader.java:74)
at org.apache.flink.util.FlinkUserCodeClassLoader.loadClass(FlinkUserCodeClassLoader.java:48)
at java.base/java.lang.ClassLoader.loadClass(Unknown Source)
```

**Producer**

I was testing HOP pipelines on a local environment with Kubernetes. I generated a HOP fat-jar and deployed it on our flink-kubernetes-operator. As a schema registry we are using apicurio-registry-sql (https://github.com/Apicurio/apicurio-registry) that is fully compatible with Confluent schema registry API.

There was an exception in the job manager while using Beam Kafka Produce for writing Avro messages to Kafka using a schema registry
Please see the attached example-kafka-avro-produce.hpl.

The exception that I got in the job manager:
```

Caused by: java.lang.NullPointerException
at org.apache.beam.sdk.coders.AvroCoder$AvroDeterminismChecker.checkIndexedRecord(AvroCoder.java:629)
at org.apache.beam.sdk.coders.AvroCoder$AvroDeterminismChecker.recurse(AvroCoder.java:497)
at org.apache.beam.sdk.coders.AvroCoder$AvroDeterminismChecker.check(AvroCoder.java:476)
at org.apache.beam.sdk.coders.AvroCoder.(AvroCoder.java:316)
at org.apache.beam.sdk.coders.AvroCoder.(AvroCoder.java:308)
at org.apache.beam.sdk.coders.AvroGenericCoder.(AvroGenericCoder.java:26)
at org.apache.beam.sdk.coders.AvroGenericCoder.of(AvroGenericCoder.java:30)
at org.apache.beam.sdk.coders.AvroCoder.of(AvroCoder.java:151)
at org.apache.hop.beam.transforms.kafka.BeamProduceMeta.handleTransform(BeamProduceMeta.java:149)
at org.apache.hop.beam.pipeline.HopPipelineMetaToBeamPipelineConverter.handleBeamOutputTransforms(HopPipelineMetaToBeamPipelineConverter.java:344)
at org.apache.hop.beam.pipeline.HopPipelineMetaToBeamPipelineConverter.createPipeline(HopPipelineMetaToBeamPipelineConverter.java:219)
```

Pipelines to reproduce
[Archive.zip](https://github.com/apache/hop/files/11002555/Archive.zip)

### Issue Priority

Priority: 2

### Issue Component

Component: Beam, Component: Transforms

Start with the attached Archive.zip pipelines and the stack traces, then inspect org.apache.hop.beam.transforms.kafka.BeamProduceMeta at line 149 and the Beam pipeline converter. Reproduce the consumer and producer failures in the stated Flink deployment; done means Avro messages can use the schema registry without the missing serializer class or AvroCoder exception.

Written by the indexing model from the issue text.