Parquet file can not be read properly when it starts with an underscore

When a Parquet file starts with an underscore,
- parquet-tools head fails
- Hive/Impala will not show any data in external tables

If the underscore is deleted, everything works as expected. I don't know if this is expected behaviour - but if so, it should probably be documented.

**Environment**: Cloudera CDH 5.10.1 (parquet-tools-1.5.0-cdh5.10.1.jar)
Mac OS Sierra 10.12.4 (parquet-tools-1.9.0.jar)
**Reporter**: [Jurgis Pods](https://issues.apache.org/jira/secure/ViewProfile.jspa?name=pederpansen)

_{**Note**: *This issue was originally created as [PARQUET-971](https://issues.apache.org/jira/browse/PARQUET-971). Please see the [migration documentation](https://issues.apache.org/jira/browse/PARQUET-2502) for further details.*}

No source file, test, or entry point is named. Start by reproducing the parquet-tools failure with a file whose name starts with an underscore, then compare behavior with Hive and Impala and determine whether the intended outcome is a code fix or documentation; done means the behavior is corrected or clearly documented.

Written by the indexing model from the issue text.