Pyusid

Minor update to previous pyUSID version with changes to ensure compatibility with updated methods of slicing of h5 datasets in h5py. This also resolves the errors thrown by visualization utilizes that have plagued BGlib.

This version maintains compatibility with the latest numpy, and also re-instates comp_utils that was removed from sidpy, causing version incompatibility. Furthermore, some dependency requirements have been updated. Whilst there are no feature improvements, this release ensures continued functionality of pyUSID.

Minor release to fix failing functions with updated versions of pillow, and a minor bug fix to reshaping circumstance where there is only one position in a spectral dataset.

**Note:** Legacy holdouts such as ``write_utils`` will be removed in the next version.

* Minor bug fixes
* Function that writes ``sidpy.Dataset`` to USID formatted HDF5 files.
* Refactoring
* Renamed ``numpy_translator`` to ``array_translator``
* Renamed ``write_utils`` to ``anc_build_utils``
* Separate module for ``Dimension``
* Removed unnecessary files like ``dtype_utils``, ``io_utils``
* Updated imports to use ``sidpy`` instead of ``pyUSID``

Major changes
---------------
* Moved USID-agnostic functions in following modules to to new package - ``sidpy`` since these will be shared with new sister package ``pyNSID``. The skeletons of the moved methods and classes will remain in pyUSID but are actually calling ``sidpy`` underneath. Deprecated local functions will be removed in a later release. Users are advised to use ``sidpy`` instead. List of affected modules:
* ``pyUSID.io``:
* ``.io_utils.py``
* ``.write_utils.py``
* ``.translator.py``
* ``.reg_ref.py``
* ``.dtype_utils.py``
* ``.hdf_utils.simple.py``,
* ``pyUSID.viz``:
* ``.plot_utils.py``
* ``.jupyter_utils.py``
* ``pyUSID.processing.comp_utils.py``

Minor changes
---------------
* bug fixes
* New "10 minutes to pyUSID document" thanks to rajgiriUW
* Fixed bug regarding dask array transpose in ``pyUSID.io.USIDataset``
* Not writing region references for ancillary datasets anymore. Region reference functionality will be completely removed from pyUSID in an upcoming release since it is not being used meaningfully anywhere.
* Enabling multiple instances of ``pyUSID.Process`` class to be executed in parallel via each MPI rank in order to facilitate ensemble / embarrassingly parallel processing.
* No longer pushing tests directory when deploying via pip. ``site-packages`` directory will no longer have tests belonging to pyUSID. Users interested in adding, evaluating pyUSID tests are recommended to git clone the GitHub repository instead

Ability to work on multiple HDF5 files
The io functions in this version check and facilitate I/O operations between multiple HDF5 files. Importantly these capabilities have been passed onto the `Process` class which now enables child classes to check for and write results into different files if required.

These upgrades are in preparation for use with [DataFed](https://datafed.ornl.gov) which is capable of capturing provenance across multiple files / records. Now, it is not necessary for results to be contained within the same HDF5 file for provenance tracking.

**Note**: Operations by default will continue to operate within the same files unless appropriate keyword arguments are passed.