Add case insensitive support to Parquet.

[rdblue]

#83 and #89 add case sensitivity flags, but filters pushed down to Parquet still do not support case sensitivity. This updates the Parquet row group filters to support a case sensitivity flag and updates Spark to pass the flag.

This also updates the old read path that uses ReadSupport when converting Iceberg expressions to Parquet filters.

[rdblue]

@xabriel, can you review? We ran into this when testing case insensitive queries.

[xabriel]

Could you also provide a repro SQL query that would trigger this issue as a comment to this PR? I guess my simple test cases didn't.

[rdblue]

@xabriel, thanks for looking. You're right that I still need to add tests.

[rdblue]

@xabriel, can you take another look? I've fixed the problem you found and validated there are tests that actually exercise case insensitive binding for the Parquet classes.

To your question about how to reproduce this, you have to run a query that pushes filters down to Parquet. The filters you use can't be satisfied by partition filtering because Iceberg will drop those filters and won't push them down. I would recommend writing a dataset without partitioning and sorted by a categorical value, then test equality filters on that categorical value. That would hit this problem.

[xabriel]

Looks good @rdblue. LGTM.

[rdblue]

Thanks for reviewing this, @xabriel! I'll merge it.