[WIP][SPARK-29228][SS] Adding the ability to add headers to the KafkaRecords published to Kafka from spark structured streaming

[jhsb25]

What changes were proposed in this pull request?

The changes in the pull request allow a developer to provide a column within their structured streaming dataframe titled header and of type map<string, string> which when the dataframe is sinked to kafka the headers will be applied to the producer record. This can have many applications but the main reason we wanted this introduced was for tracing purposes.

Why are the changes needed?

Sometimes it is necessary to add custom headers to the kafka producer record but with the current spark-kafka sink this is not possible, these changes make it optional to add headers to the kafka records that will be produced.

Does this PR introduce any user-facing change?

If the user wants to apply headers in the records that will be published to kafka they can provide a new column of type map<string, string> with the title header:

df.selectExpr("topic", "CAST(key AS STRING)", "CAST(value AS STRING)", "map('custom-header-1', 'foo', 'custom-header-2', 'bar') as headers")
  .write
  .format("kafka")
  .option("kafka.bootstrap.servers", "host1:port1,host2:port2")
  .save()

How was this patch tested?

Unit tests have been added and it has been tested thoroughly on a cluster.

[AmplabJenkins]

Can one of the admins verify this patch?