Perform encoding before bloom filters write out

When entropy of bloom filter is low say nearly all 0 or all 1, we may want to consider to use encoding to save space.

**Reporter**: [Junjie Chen](https://issues.apache.org/jira/secure/ViewProfile.jspa?name=junjie) / @chenjunjiedada

_{**Note**: *This issue was originally created as [PARQUET-1495](https://issues.apache.org/jira/browse/PARQUET-1495). Please see the [migration documentation](https://issues.apache.org/jira/browse/PARQUET-2502) for further details.*}

The issue names no files, tests, or entry points. Begin by locating the bloom-filter write path and the existing encoding-related tests, then determine the expected behavior for low-entropy filters. Done would require an agreed encoding approach and coverage showing that bloom filters are written correctly and can be read back.

Written by the indexing model from the issue text.