Flatten-dict

Change
- require `importlib-metadata` for Python<3.8 to improve importing performance (46)
- support enumerating the objects that don't support `bool()` (42, 49)

Code Quality
- migrate to Github Actions and test Python 3.10 (47, 48)

Change
- improve import speed when Python >= 3.8 or `importlib-metatdata` is installed (44) (thanks isidentical)

New
- add `max_flatten_depth` to `flatten()` (22, 31, 35)

Change
- made `pathlib2` optional for Python < 3.4 (32)
- use `poetry-core` instead of `poetry` in `pyproject.toml` (36)
- deprecate `flatten_dict.splitter` in favor of `flatten_dict.splitters` (38)
- deprecate `flatten_dict.reducer` in favor of `flatten_dict.reducers` (38)

Code Quality
- improve code quality (32)
- test for Python 3.9 (34)

New
- add `dot` splitter and reducer (20) (thanks zemelLeong)
- add factory functions `make_reducer()` and `make_splitter` to create reducer and splitter with delimiter (25) (thanks hsorsky)

Fix
- cast type to str when using `underscore` reducer (21) (thanks gllrt)

New
- add `underscore` splitter and reducer (13) (thanks duarteocarmo)
- add `keep_empty_types` to `flatten()` (18)

New
- add a parameter `enumerate_types` to `flatten()` (5, 7)