Aioinflux

- Fix bug where timezone-unaware datetime input was assumed to be in local time (11 / a8c81b7)
- Minor improvement in dataframe parsing (1e33b92)

Highlights:

- Drop Pandas/Numpy requirement (9)
- Improved iteration support (816a722)
- - Implement tag/key value caching (9a65787)
- Improve dataframe serialization
- Speed improvements (ddc9ecc)
- Memory usage improvements (a2b58bd)
- Disable concatenating of dataframes of the same measurement when grouping by tag (331a0c9)
- Queries now return tag columns with `pd.Categorical` dtype (efdea98)
- Writes now automatically identify `pd.Categorical` dtype columns as tag columns (ddc9ecc)

API changes:

- `mode` attribute was "split" into `mode` and `output`.
Default behavior remains the same (async / raw).
- Iteration is now made easier through the `iterable` mode
and `InfluxDBResult` and `InfluxDBChunkedResult` classes

Highlights

- Documentation is now complete
- Improved iteration support (via `iter_resp`) (cfffbf5)
- Allow users to add custom query patterns
- Add support for positional arguments in query patterns
- Reimplement `__del__` (40d0a69 / 7)
- Improve/debug dataframe parsing (7beeb53 / 96d78a4)
- Improve write error message (7972946) (by miracle2k)

API changes:
- Rename `AsyncInfluxDBClient` to `InfluxDBClient` (54d98c9)
- Change return format of chunked responses (related: cfffbf5 / 6)
- Make some `__init__` arguments keyword-only (5d2edf6)

- Add `__aenter__`/`__aexit__` support (5736446) (by Kargathia)
- Add HTTPS URL support (49b8e89) (by miracle2k)
- Add Unix socket support (8a8b069) (by carlos-jenkins)
- Fix bug where tags where not being added to DataFrames when querying (a9f1d82)

- Add error handling for chunked responses (db93c20)
- Fix DataFrame tag parsing bug (aa02faa)
- Fix boolean field parsing bug (4c2bff9)
- Increase test coverage

Initial release.
The API is relatively stable but there might be some bugs here and there.
Discretion advised when using in production.