Add support to write DataFrames to an object store

**Is your feature request related to a problem or challenge? Please describe what you are trying to do.**
Hey folks :wave: we're evaluating Datafusion/Ballista to perform queries as part of a pipeline on some pretty large datasets (we're talking on the order of a few Terabytes at minimum) and would like to read from Parquet files from GCS and write as Parquet to a different GCS bucket. Right now it's possible to get a DataFrame from GCS using the object_store crate, but it's not possible to write the resulting DataFrame back to GCS.

**Additional context**
This feature was already discussed in #2185 but that issue is out of date because the new object_store crate has since been merged in.

Start by reviewing the existing object_store crate integration for reading Parquet files from GCS, then trace how a DataFrame is currently written to Parquet. Define the expected behavior for writing the resulting DataFrame to a different GCS bucket and verify it with an end-to-end object-store case.

Written by the indexing model from the issue text.