[Data] Make Parquet tests more robust and expose Parquet logic

[bveeramani]

Why are these changes needed?

This PR makes the following changes:

• Rewrites test_dataset_stats_read_parquet as test_dataset_stats_range to make the test independent of the Parquet implementation (the intent of the test isn't Parquet-specific)
• Makes test_parquet_read_spread robust to different Parquet implementations (the test makes assumptions about how tasks are launched)
• Makes test_fsspec_filesystem not depend on the number of files written by Ray Data (the test assumes write_parquet writes exactly two files)
• Exposes Parquet-specific logic to other modules
  • Removes underscores from SerializedFragment and check_for_legacy_tensor_type
  • Refactors ParquetDatasource._sample_fragments as sample_fragments
  • Introduces get_parquet_dataset

Related issue number

Checks

• I've signed off every commit(by using the -s flag, i.e., git commit -s) in this PR.
• I've run scripts/format.sh to lint the changes in this PR.
• I've included any doc changes needed for https://docs.ray.io/en/master/.
  • I've added any new APIs to the API Reference. For example, if I added a
    method in Tune, I've added it in doc/source/tune/api/ under the
    corresponding .rst file.
• I've made sure the tests are passing. Note that there might be a few flaky tests, see the recent failures at https://flakey-tests.ray.io/
• Testing Strategy
  • Unit tests
  • Release tests
  • This PR is not tested :(