Frictionless

- Initial API/codebase improvements for v4 (BREAKING)
- Allow `Package/Resource(source)` notation (guess descriptor/path/etc)
- Renamed `schema.infer` -> `Schema.from_sample`
- Renamed `resource.inline` -> `resource.memory`
- Renamed `compression_path` -> `innerpath`
- Renamed `compression: no` -> `compression: ""`
- Updated `Package/Resource.infer` not to infer stats (use `stats=True`)
- Removed `Package/Resource.infer(only_sample)` argument
- Removed `Resouce.from/to_zip` (use `Package.from/to_zip`)
- Removed `Resouce.source` (use `Resource.data` or `Resource.fullpath`)
- Removed `package/resource.infer(source)` argument (use constructors)
- Added some new API (will be covered in the updated docs after the v4 release)

- Make Resource independent from Table/File (607) (BREAKING)
- Resource can be opened like Table (it's recommended to use Resource instead of Table)
- Renamed `resource.read_sample()` to `resource.sample`
- Renamed `resource.read_header()` to `resource.header`
- Renamed `resource.read_stats()` to `resource.stats`
- Removed `resource.to_table()`
- Removed `resource.to_file()`

- Optimize Row/Header/Table and rename header errors (601) (BREAKING)
- Row object is now lazy; it casts data on-demand preserving the same API
- Method `resource/table.read_data(_stream)` now includes a header row if present
- Renamed `errors.ExtraHeaderError->ExtraLabelError` (`extra-label-error`)
- Renamed `errors.MissingHeaderError->MissingLabelError` (`missing-label-error`)
- Renamed `errors.BlankHeaderError->BlankLabelError` (`blank-label-error`)
- Renamed `errors.DuplicateHeaderError->DuplicateLabelError` (`duplicate-label-error`)
- Renamed `errors.NonMatchingHeaderError->IncorrectLabelError` (`incorrect-label-error`)
- Renamed `schema.read/write_data->read/write_cells`

- Renamed aws plugin to s3 (594) (BREAKING)

bash
$ pip install frictionless[aws] before
$ pip install frictionless[s3] after

- Drafted support for writing Multipart Data (583)

- Added support for writing to Remote Data (582)