Reduce memory pressure when reading footers

I encountered OOMs reading metadata for a dataset with 500+ columns, many of them quite sparse.

We can reduce memory utilization significantly.

**Reporter**: [Dmitriy V. Ryaboy](https://issues.apache.org/jira/secure/ViewProfile.jspa?name=dvryaboy) / @dvryaboy

_{**Note**: *This issue was originally created as [PARQUET-11](https://issues.apache.org/jira/browse/PARQUET-11). Please see the [migration documentation](https://issues.apache.org/jira/browse/PARQUET-2502) for further details.*}