Etlhelper

This minor release includes:

+ Recipe for [database to API transfer](https://github.com/BritishGeologicalSurvey/etlhelper#database-to-api--nosql-copy-etl-script-template) uses asynchronous `aiohttp` for ~10x speedup.
+ `iter_chunks`, `fetchmany`, `fetchone` close transactions when they finish.
+ README describes using `fetchone` with [queries using `SELECT ... RETURNING`](https://github.com/BritishGeologicalSurvey/etlhelper#insert-rows) syntax.

This minor release includes:

+ Use `fast_executemany` by default for MS SQL Server (via PyODBC driver). This provided a 100x increase in insert speed during our tests. See [README.md](https://github.com/BritishGeologicalSurvey/etlhelper#disabling-fast_executemany-for-sql-server-and-other-pyodbc-connections) for details.

This minor release includes:

+ allow setting of `row_factory` in `copy_rows`. Thanks to Real-Gecko.
+ basic database to API ETL pipeline script added as recipe in README.md

This minor release includes:

+ `setup_oracle_client` fails gracefully with sys.exit() instead of producing stack trace
+ README highlights `iter_rows`
+ Fix typos in README example ETL script

This minor release includes:

- Wheel distribution built
- setup_oracle_client script works with `sh` and `zsh`
- README explains inplace dictionary transforms

This minor release improves error messages related to DbParams and connections.

+ Creating DbParams for database type with no installed driver raises a Warning
+ Connecting to database with no installed driver raises ETLHelperConnectionError
+ Stack trace for non-existent dbtype error simplified
+ `sys.exit(1)` / SystemExit calls replaced with ETLHelper errors
+ All print statements replaced with logging calls - this keeps `stdout` clean