feat: Add KafkaRecord type and Protobuf deserialization for Kafka trigger binding (#868)

[TsuyoshiUshio]

Summary

Add support for binding to raw Apache Kafka records (KafkaRecord type) in the Java worker, enabling users to access full Kafka message metadata via Protobuf deserialization.

Issue

Fixes #868
Relates to Azure/azure-functions-kafka-extension#612

Changes

New files

• KafkaRecord.java — Main POJO with topic, partition, offset, key/value as raw byte[], timestamp, headers, leader epoch
• KafkaHeader.java — Header with key (String) + value (byte[]) + getValueAsString() helper
• KafkaTimestamp.java — Timestamp with getUnixTimestampMs(), getType(), getDateTimeOffset()
• KafkaTimestampType.java — Enum: NotAvailable, CreateTime, LogAppendTime (with fromValue() safe mapping)
• KafkaRecordProtoDeserializer.java — Protobuf → KafkaRecord POJO conversion
• KafkaRecordProto.proto — Shared schema (synced with host extension Microsoft.Azure.WebJobs.Extensions.Kafka)
• KafkaRecordDeserializerTest.java — 10 unit tests

Modified files

• RpcModelBindingDataSource.java — Add content_type dispatch: application/json (existing path) vs application/x-protobuf with source AzureKafkaRecord (new KafkaRecord path). Also adds KafkaRecord.class operation to DataOperations
• pom.xml — Add second protobuf-maven-plugin execution for Kafka proto compilation from src/main/proto/kafka/

Breaking Changes

None. All existing binding types (String, byte[], Map<String, String>, POJO) continue to work unchanged. Users opt in via KafkaRecord parameter type.

Testing

• 10 unit tests covering: full record round-trip, null key/value, no leader epoch, unknown timestamp type fallback, multiple headers, null header value, timestamp OffsetDateTime, KafkaTimestampType.fromValue()
• E2E tests: deferred (requires Docker Kafka environment)

User Experience

// Existing (continues to work)
@FunctionName("ExistingTrigger")
public void run(@KafkaTrigger(...) String message) { }

// NEW: Full record access
@FunctionName("KafkaRecordTrigger")
public void run(
    @KafkaTrigger(name = "record", topic = "my-topic",
                  brokerList = "%BrokerList%", consumerGroup = "$Default")
    KafkaRecord record,
    final ExecutionContext context) {
    context.getLogger().info("Topic: " + record.getTopic());
    context.getLogger().info("Key: " + record.getKeyAsString());
    for (KafkaHeader h : record.getHeaders()) {
        context.getLogger().info("Header: " + h.getKey() + " = " + h.getValueAsString());
    }
}

Note

The host-side counterpart (Kafka Extension 4.3.1) is already released on NuGet. The .NET Isolated Worker implementation was completed in Azure/azure-functions-dotnet-worker#3356.

[TsuyoshiUshio]

Dependency note: This PR depends on Azure/azure-functions-java-library#236 which adds the KafkaRecord, KafkaHeader, KafkaTimestamp, KafkaTimestampType POJO types to the library.

The POJO types were initially in this PR but have been moved to the library so that user function code can import them. This PR now contains only the runtime-side changes:

• KafkaRecordProtoDeserializer — Protobuf → KafkaRecord conversion
• RpcModelBindingDataSource — content_type dispatch (JSON vs Protobuf)
• Proto schema and Maven plugin config

Release order: java-library must be released first → update worker's java-library dependency version → merge this PR.