Update the Python package metadata and dataset loader

[jjacobson95]

Pipeline Hardening/Debugging PR # 15

Update the Python package metadata and dataset loader

Brings the installable coderdata Python package in line with the new datasets and the new phosphoproteomics data type, so users can download and load everything the build now produces.

Dataset registry (coderdata/dataset.yml)

• Bumps the dataset version to 2.3 and points at the corresponding Figshare article for downloads.
• Registers the cnf dataset (cutaneous neurofibroma patient-derived organoid drug screen) with its description, phosphoproteomics data type, and source citation, so the package can download and load it when it is available.

Loader (coderdata/dataset/dataset.py)

• Adds phosphoproteomics and phosphosites as first-class loadable attributes on the dataset object, each with the usual getter, setter, and deleter, so the new data type loads alongside transcriptomics, proteomics, and the rest.

Downloader (coderdata/download/downloader.py)

• Recognizes phosphosites files when selecting what to download, so the phosphosite reference is fetched with the dataset when it is available.

Notes

• This brings the package in step with the current published release (version 2.3). It is a catch-up update: the package will need another update for version 2.4, which is expected to be released soon.
• Support for the cNF dataset and the phosphosites reference (registration, download, and load) is included in the package so both work as soon as they are available.

Scope: 3 files. Base: validation-and-improve-prep. This is the top of the stack.

Related issues

Closes #477

[ymahlich]

Looks good.