Pynonymizer

Added
- `--workers` option to control number of threads to use when anonymizing. Table anonymization will be done in threads and should result in a major speed boost. By default, a single worker is used.
- `--mssql-ansi-warnings-off` option (default: on) to disable ansi warnings when making updates

Fixed
- Fixed a bug where using the raw `pynonymizer` command would cause an unexpected error.

Added
- Better error messages for unsupported fake types - the error should now explain the problem and link to the docs in the right section. [133]
- Error message for when a fake_type is used with the wrong config kwargs (these would have previously been caught under "unsupported fake types")
- event hooks for progress events. you can now use your own progress bar if you're invoking the process via the python interface.
- support for connection strings in mssql.

Changed
- removed credentials validation (db_user, db_password) from the initial cli validation, as these are now handled by your original database. This should grant the most flexibility for different types of credentials.

Removed
- Positional INPUT. Use the -i/--input option instead
- Positional STRATEGYFILE. Use the -s/--strategy option instead
- Positional OUTPUT. Use the -o/--output option instead
- Deprecated environment keys: `DB_TYPE, DB_HOST, DB_NAME, DB_USER, DB_PASS`.
- `--fake-locale` `-l` `$PYNONYMIZER_FAKE_LOCALE` cli option. Use the `locale:` key in your strategyfile instead
- `dotenv` is no longer a dependency. you are expected to load your environment variables yourself!
- Removed local server check for mssql, as manual overrides using connection string will be possible. pynonymizer isn't going to stop you!

Changed
- Postgres FAKE_UPDATE strategy now uses an id-based randomization to pick from the seed table.
This speeds up updates on large tables, as the seed table no longer needs to be ordered for every row in the target table.
Also this makes performances less dependent on the length of the seed table.

Changed
- In MySQL,`db_host`, `db_user`, `db_pass` and `db_port` are now optional and will be omitted from CLI options if not passed.
You can now use the `my.cnf` file to pass config to the mysql CLI. [108]. Read more about the rationale behind these changes in [doc/database-credentials.md](doc/database-credentials.md).
- In Postgres, `db_pass` will be optional and will be omitted from CLI options if not passed. You can now use the `.pgpass` file to securely pass the password to the pg CLI. Read more about the rationale behind these changes in [doc/database-credentials.md](doc/database-credentials.md).

Added
- Added offical test support for python 3.10
- Added a new option, `--ignore-anonymization-errors` that will allow the anonymization step to error without propagating errors upstream. This is useful if you always want the resulting dumpfile, even when there are db or schema faults.
Changed
- Changed how seed rows are assigned their default value and made sure this value made it's way into the documentation (150 rows by default).
Deprecated
- Column Strategy "empty" has been deprecated because its effect was inconsistent on different providers and column types.
using it will generate a warning on config parse. It has been removed from the documentation so as to cut down on confusion.
The recommended way to update a column to empty is to use a `literal` set to the appropriate "empty" data for the column.
Removed
- Removed offical test support for python 3.6
- Removed process step `GET_SOURCE`(no-op) as it was causing confusion. This is not considered a breaking change as it was never implemented.

Changed
- Changed anonymization process to attempt to anonymize all tables before throwing errors. [96]
Fixed
- Fixed a bug in mysql/postgres that didn't wait for the restore dump process to complete before starting the anonymize procedure