Use Spark DataSourceV2 to handle Parquet files

[Aryex]

Descriptions

The connector support Parquet files by reusing some of Spark's lower-level internal systems. This resulted in the connector having to copy over private codes, handle data partitioning, and overall longer codes to maintain.

With Spark 3.0.0 adding support for Parquet DataSourceV2, it could be re-used to handle Parquet files like how JSON was supported in #370. Note that we would still need to look into how writing would be handled.

This could potentially be looked into as part of #403.

Reason: This change will help reduce effort supporting future Spark versions.

[alexey-temnikov]

Lowing priority, as it is relevant only during Spark Upgrade (when spark APIs are changed)