Pig and Hive cannot read repeated groups written with parquet-protobuf

parquet-hive and parquet-pig make assumptions about list schemas that are not compatible with the more compact schemas generated by parquet-protobuf. This bug was discussed in more detail on https://github.com/Parquet/parquet-mr/issues/354

**Reporter**: [Nathan Howell](https://issues.apache.org/jira/secure/ViewProfile.jspa?name=NathanHowell)

_{**Note**: *This issue was originally created as [PARQUET-14](https://issues.apache.org/jira/browse/PARQUET-14). Please see the [migration documentation](https://issues.apache.org/jira/browse/PARQUET-2502) for further details.*}

首先阅读 parquet-hive 和 parquet-pig 中对 list-schema 的处理，然后利用 issue 354 中的讨论来了解 parquet-protobuf 如何生成紧凑的 repeated-group schema。完成的标准是两个 reader 都能接受这些 schema，同时不破坏现有的 list-schema 处理；内容没有指定具体的文件或测试，因此必须先定位这些入口点。

由索引模型根据 Issue 内容生成。