refine_models step

[ilibarra]

• [x] implement notebook refine_models.ipynb that takes a unified model from the aggregate step, all the available data in a concatenated way (see concatenation and sparse on/off), and applies a refinement of the activities/etas plus soft update of the filter weights.
• [ ] adapt refine.ipynb into a refine_filters.py, to be used in the pipeline with the current output.

[ilibarra]

template notebook below. Now duplicate of #20 https://github.com/theislab/mubind/blob/development/notebooks/concatenation/01_load_and_concatenate.ipynb