feat: add read/write support

[henryiii]

Adding read/write support, hdf5 support pulled from scikit-hep/boost-histogram#997. Zip and JSON support added as well. Libraries supporting the UHI schema can simply produce or consume a Python object following the schema and then use the utilities in uhi.io to read or write to all the formats we implement.

This also makes a clarification to the schema:

• The storage format needs to support ND Arrays, so that's explicitly shown as nested lists now in the schema. For interlibrary communication, it must support np.asarray.

Each format remains as flexible as possible by expecting the user to do a little setup; you can store anywhere in an HDF5 file via groups. You can write to an existing zip file, control the compression, etc. by using a Zipfile object.

I'm using the JSON format to store tests, then converting them to the binary formats (HDF5 and Zip) and then testing that the files are correct and can be read back.

[APN-Pucky]

I just want to drop a comment and I am not very familiar with hdf5: YODA-2.1.0 supports hdf5 since the latest release https://yoda.hepforge.org/. For me it will be interesting if/how the UHI-hdf5 interoperates with YODA-hdf5 as part of https://github.com/APN-Pucky/babyyoda.