nested column support for Parquet and Avro

[clintropolis]

Description

This PR officially adds support for Parquet and Avro for Druid nested columns, including using the 'JSON' transformation functions at ingestion time. Along the way, fixes a couple of issues with using these same functions during ingest time, particularly when the top level object is a List/Array instead of an object. This particular code area was a bit obsolete, expecting only the primitive Druid ARRAY types to be handled, but that is no longer true after the introduction of nested arrays and complex expression types (e.g. arrays of complex types). I've expanded this code area to fall back on using ExprEval.bestEffortOf to determine the type of the array. This isn't actually specific to Parquet or Avro, these errors could occur with JSON and potentially any other nested list input values intended to be used in expressions.

Release note

Druid nested columns now support Parquet and Avro input formats, either to ingest directly into nested columns, or using the 'JSON' expressions as transforms, such as JSON_VALUE and JSON_QUERY.

---

This PR has:

• been self-reviewed.
• added documentation for new or modified features or behaviors.
• a release note entry in the PR description.
• added Javadocs for most classes and all non-trivial methods. Linked related entities via Javadoc links.
• added comments explaining the "why" and the intent of the code wherever would not be obvious for an unfamiliar reader.
• added unit tests or modified existing tests to cover new code paths, ensuring the threshold for code coverage is met.
• been tested in a test Druid cluster.