Jdata

* support [blosc2](https://www.blosc.org/) family meta-compressors (`blosc2blosclz`, `blosc2lz4`, `blosc2lz4hc`, `blosc2zlib`, `blosc2zstd`)
* optimized numpy array encoding/decoding to avoid overhead
* code beautification with [black](https://pypi.org/project/black/)
* return undecoded data when a codec is not installed
* add [benchmark](https://mail.python.org/archives/list/numpy-discussionpython.org/message/TUO7CKTQL2GGH2MIWSBH6YCO3GX4AV2O/) comparing various compressors for saving/loading numpy arrays

- support `lzma` compressed data
- encode/decode numpy complex-valued scalars, 2
- add unit tests

ChangeLog:

- migrate testing script to use unittest