Datar

- ⬆️ Bump simplug to 0.2.2
- ✨ Add `apis.other.array_ufunc` to support numpy ufuncs
- 💥 Change hook `data_api` to `load_dataset`
- ✨ Allow backend for `c[]`
- ✨ Add `DatarOperator.with_backend()` to select backend for operators
- ✅ Add tests
- 📝 Update docs for backend supports

- 🚑 Fix false warning when importing from all

- Pump simplug to 0.2

- Detach backend support, so that more backends can be supported easier in the future
- numpy backend: https://github.com/pwwang/datar-numpy
- pandas backend: https://github.com/pwwang/datar-pandas
- Adopt pipda 0.10 so that functions can be pipeable (148)
- Support pandas 1.5+ (148), but v1.5.0 excluded (see pandas-dev/pandas48645)

- Pump pipda to 0.8.0 (fixes 149)

Fixes

- Fix `weighted_mean` not handling group variables with NaN values (137)
- Fix `weighted_mean` on `NA` raising error instead of returning `NA` (139)
- Fix pandas `.groupby()` used internally not inheriting `sort`, `dropna` and `observed` (138, 142)
- Fix `mutate/summarise` not counting references inside function as used for `_keep` `"used"/"unused"`
- Fix metadata `_datar` of nested `TibbleGrouped` not frozen

Breaking changes

- Refactor `core.factory.func_factory()` (140)
- Use `base.c[...]` for range short cut, instead of `f[...]`
- Use `tibble.fibble()` when constructing `Tibble` inside a verb, instead of `tibble.tibble()`
- Make `n` a keyword-only argument for `base.ntile`

Deprecation

- Deprecate `verb_factory`, use `register_verb` from `pipda` instead
- Deprecate `base.data_context`

Dependences

- Adopt `pipda` `v0.7.1`
- Remove `varname` dependency
- Install `pdtypes` by default