Populate sources refactor

The `populate_sources` method of the observatory is used to re-create datasets from raw data files on disk.

Since this is a side-utility, and not part of the main analysis, it may be possible to achieve the same result using direct kwargs input, instead of through the `Parameters` object.

Also, we should consider if the right thing to do is "populate_raw_data" instead of "populate_sources". It may be something that combines these two.