Add PARQUET_1_0 and non-repeated data performance tests to parquet-benchmarks

The current parquet-benchmarks module run some performance tests between different block & page sizes for PARQUET_2_0 version only. We should run some tests with PARQUET_1_0 version as well in order to get a view about new parquet version enhancements, and be able to catch possible overheads early by comparing with the old file format.

Also, this module uses repeated data to benchmark the settings. We should also use random data to get different results about how current and new encodings work with real world data.

**Reporter**: [Sergio Peña](https://issues.apache.org/jira/secure/ViewProfile.jspa?name=spena) / @spena
**Assignee**: [Sergio Peña](https://issues.apache.org/jira/secure/ViewProfile.jspa?name=spena) / @spena
#### PRs and other links:
- [PR-224](https://github.com/apache/parquet-mr/pull/224)

_{**Note**: *This issue was originally created as [PARQUET-315](https://issues.apache.org/jira/browse/PARQUET-315). Please see the [migration documentation](https://issues.apache.org/jira/browse/PARQUET-2502) for further details.*}

Start in the parquet-benchmarks module and review the existing PARQUET_2_0 performance tests that vary block and page sizes. Add coverage for PARQUET_1_0 and random data alongside the repeated-data benchmarks, then compare the results to confirm both format versions and data patterns are exercised; check PR-224 for prior work and context.

Written by the indexing model from the issue text.