Add Spark-compatible Parquet reader

### Is your feature request related to a problem or challenge?

I would like to add a Spark-compatible Parquet reader to the `datafusion-spark` crate, based on functionally that already exists in Comet.

The Parquet reader should support per-Spark-version behavior. For example, there were significant differences between Spark 3.x and 4.x with regards to type widening and handling of `TimestampNTZ`.

### Describe the solution you'd like

_No response_

### Describe alternatives you've considered

_No response_

### Additional context

_No response_

Start by reading the datafusion-spark crate and locating the existing Comet functionality referenced in the issue. Compare the Spark 3.x and 4.x behavior for type widening and TimestampNTZ, then define done as a Parquet reader that applies the correct behavior for each supported Spark version.

Written by the indexing model from the issue text.