Refactor!(hive): improve transpilation of TO_JSON

[georgesittas]

Related discussion: #1803 (comment)

This is a best-effort approach, so it might not be complete. The difficulty of transpiling these JSON calls correctly stems from the fact that Hive / Spark work with JSON values using only their string representation (*), whereas Trino and Presto have native support for them, using a JSON type.

Note: all changed tests were previously failing in Spark. The new ones work and have the same semantics as in Trino.

(*) Or by parsing them, e.g. with FROM_JSON, and converting them into native struct types like struct and map.

References:

• https://trino.io/docs/current/functions/json.html
• https://spark.apache.org/docs/latest/api/sql/index.html#from_json
• https://spark.apache.org/docs/latest/api/sql/index.html#schema_of_json
• https://spark.apache.org/docs/latest/api/sql/index.html#to_json
• https://spark.apache.org/docs/latest/api/sql/index.html#regexp_extract