Clean up code around create_sorting_analyzer and create/load_* functions in sortinganalyzer.py

[ecobost]

This PR does not functionally change much. It is mainly for code refactoring and to improve legibility.

The only two functional changes are:

• fix small bug when calling save_to_zarr("test.zarr") would add an extra .zarr to the filename (

  spikeinterface/src/spikeinterface/core/base.py

  Line 1087 in 39dfac4

  zarr_path = cache_folder / f"{name}.zarr"

  )
• when return_scaled and return_in_uV are provided, prioritize return_in_uV. Currently, it prefers return_scaled over return_in_uV

Code quality improvements:

• type hinting
• Collect code into logical blocks (e..g, everything that has to do wit main indices together); add comment headers (e.g., Saving main indices) and reorder this blocks (e.g., expected stuff like sorting should be saved and loaded before any of the optional things like sparsity)
• Make create_binary_folder parallel create_zarr (you can now see how they are both doing exactly the same).
• Make load_binary_folder parallel load_zarr.
• Preserve the same order of operations when loading and saving. If sorting was saved first, then load first and so on.
• Tighten code when possible

[zm711]

A few initial comments.

[alejoe91]

Thanks @ecobost

Much cleaner :) I have a few suggestions/comments. Could you tackle them? Then this is good to go for me!

[ecobost]

@alejoe91 Couple of comments still open. Let me know when this version is ok to fix merge issues (don't wanna pollute current version)

[alejoe91]

@ecobost replied to the standing comments! Let's continue and fix the conflicts

[chrishalcrow]

This is fine by me. Would be great to propagate the PeakSign and PeakMode typing across the codebase. Would fix some incorrect typing (

spikeinterface/src/spikeinterface/core/sparsity.py

Line 782 in ecf5c6d

amplitude_mode: Literal["extremum", "peak_to_peak"] = "extremum",

) !! Can do that in a follow up.