Host minimal example dataset for use in tests and examples

[Robinlovelace]

We can host:

• A minimal gzipped csv
• Parquet translation

As a release, e.g.: https://github.com/Robinlovelace/spanishoddata/releases/tag/v0.0.1

If that does not allow direct querying of the parquet files, as I expect, we can host elsewhere e.g. on GitHub pages.

[e-kotov]

Uploading to releases works (at lest for not so large data sets) as documented in this brilliant example: https://docs.ropensci.org/piggyback/articles/cloud_native.html#duckdb. The files in a GitHub release can be queried efficiently using {duckdb}. However, this is of course inferior to just having the parquet files in a hive style format on S3 storage. But probably good enough for demonstrating proof-of-concept.

[Robinlovelace]

That is really good to see!