ParquetReader directory read using InputFile

Using Hadoop Path, users can read from directories (excluding hidden files) using a single ParquetReader. This is not yet possible using the InputFile API. To resolve this, a generic way should be added for directories to be read, consistent with the behaviour of the read API when using a Hadoop Path.

**Reporter**: [Atour Mousavi Gourabi](https://issues.apache.org/jira/secure/ViewProfile.jspa?name=amousavigourabi) / @amousavigourabi

_{**Note**: *This issue was originally created as [PARQUET-2404](https://issues.apache.org/jira/browse/PARQUET-2404). Please see the [migration documentation](https://issues.apache.org/jira/browse/PARQUET-2502) for further details.*}

Start by comparing the existing Hadoop Path directory-reading behavior with the InputFile API described in the issue. Define how a generic directory read should work, including excluding hidden files and preserving single-reader behavior, then verify the result against the existing read API behavior.

Written by the indexing model from the issue text.