An effort at dataset docstring standardization

[calebrob6]

I've noticed that the docstrings of some datasets aren't complete (in the sense that an unfamiliar user wouldn't know what the data looks like even after reading the docstring). We should go through the docstrings of existing datasets and see if we can standardize some data/formatting. I think including the following for the non-geospatial datasets would be a good start:

• Sensor
• Task
• Source of imagery
• Number of samples by split
• Size of sample
• Spatial resolution
• Number / type of bands

[adamjstewart]

I wonder how much of this is covered by our dataset tables: https://torchgeo.readthedocs.io/en/stable/api/datasets.html#non-geospatial-datasets

Without a custom ruff plugin to check for this, I think it's going to be impossible to enforce, even for me.