Datar

- 🐛 Fix `rep(df, n)` producing a nested df
- 🐛 Fix `TibbleGrouped.__getitem__()` not keeping grouping structures

- Adopt `pipda` 0.5.7
- Reimplement the split-apply-combine rule to solve all performance issues
- Drop support for pandas v1.2, require pandas v1.3+
- Remove all `base0_` options and all indices are now 0-based, except `base.seq()`, ranks and their variants
- Remove messy type annotations for now, will add them back in the future
- Move implementation of data type display for frames in terminal and notebook to `pdtypes` package
- Change all arguments end with "_" to arguments start with it to avoid confusion
- Move module `datar.stats` to `datar.base.stats`
- Default all `na_rm` arguments to `True`
- Rename all `ptype` arguments for `tidyr` verbs into `dtypes`

See more changes: https://pwwang.github.io/datar/CHANGELOG/#060

- 🐛 Hotfix for types registered for `base.proportions` (77)
- 👽️ Fix for pandas 1.4

- deps: bump pdtypes to 0.2.3 (fixing pwwang/datar208)
- fix: use `_grouper` instead of `grouper` to dismiss depr warnings starting pandas v2.2
- ci: update actions in CI

**Full Changelog**: https://github.com/pwwang/datar-pandas/compare/0.5.4...0.5.5

- deps: bump pdtypes to 0.2.2
- deps: bump pandas to 2.2.0
- chore: hide warnings for downcast in case_when (pandas 2.2+)
- chore: fix deprecation of construct Tibble from manager (pandas 2.2+)
- test: fix union not keeping original key order
- test: add tests for array_ufunc
- fix: fix future.no_silent_downcasting not existing in options of pandas < 2.2

**Full Changelog**: https://github.com/pwwang/datar-pandas/compare/0.5.3...0.5.4

- Bump datar to 0.15.3 to make np.nanmean and alike work on groupby (pwwang/datar65)

**Full Changelog**: https://github.com/pwwang/datar-pandas/compare/0.5.2...0.5.3