Add a new API in `PageReadStore` to return row ranges directly

**Environment**: Currently `PageReadStore.getRowIndexes` convert row ranges into row indexes and return that, however for some use cases (e.g., Spark) it's more efficient to operate directly on row ranges. Therefore, it may be useful to have another API in `PageReadStore` to return row ranges instead.
**Reporter**: [Chao Sun](https://issues.apache.org/jira/secure/ViewProfile.jspa?name=csun) / @sunchao

_{**Note**: *This issue was originally created as [PARQUET-2061](https://issues.apache.org/jira/browse/PARQUET-2061). Please see the [migration documentation](https://issues.apache.org/jira/browse/PARQUET-2502) for further details.*}

Start by locating PageReadStore and its getRowIndexes entry point. Read how row ranges are currently converted to row indexes, then determine the public API shape needed to return the ranges directly. Done means callers such as Spark can obtain row ranges without that conversion.

索引モデルが issue の本文から書いたものです。