[WIP] v0.3 tracking branch

[iancze]

Reopening this tracking issue for v0.3 after the first wave of commits were merged into main with #243. See #243 for more information on the scope of this reorganization.

Work introduced into this branch by #247

• closes log_likelihood loss incorrect and other loss names misleading #237
• closes TV/TSV regularizer is not equivalent to source #153
• closes Hermitian pairs of visibilities for imaging and fitting #100

Work introduced by #249

• closes Base unit for spatial frequency #223
• closes mpol.utils.convert_baselines and mpol.utils.broadcast_and_convert_baselines duplicate visread functionality #227
• Restructured mock datasets to be computed on-the-fly, using NuFFT and some reference baselines. Deprecated much of mpoldatasets and Zenodo upload architecture, overly complicated to maintain.

Remaining work:
Changes to introduce Stochastic Gradient Descent workflow

• Reorganize SimpleNet and the entire GriddedDataset workflow as an alternative option to SGD, with documentation about the use cases when one or the other might be preferred. closes New quickstart and examples #251
• Documentation/"Getting Started" to explain how MPoL library usage follows the the normal ML paradigm with SGD (closes Framing model.forward more like ML libraries? #188)
• SGD uses TensorDataset and a DataLoader closes Redesign UVDataset with Pytorch idioms in mind #162

Further documentation changes

• in the process of moving many examples to MPoL-dev/examples so they are easier to maintain
• Think about what content makes sense as rendered docs (code demonstrating key library functionality, building blocks, preferably concise) or as a longer .py file in examples/ (actual workflows following official pytorch/examples. E.g.,
  • Direct longer content to examples
    • HD 143006 imaging tutorial overview #25
    • Start examples directly from DSHARP datasets, removing need to download processed datasets from Zenodo & demonstrates required CASA work to user