Adding motion_cleaner into High-level API correct_motion

[chiyu1203]

Dear Spikeinterface community,
I bumped into the "DREDge hallucination" issue, which was described in the past #4066. Notably, when I used kilosort-like method, I did not see such spikes.
In my dataset with DREDge method, I have tried increasing win_scale, histogram_depth_smooth_um, histogram_time_smooth_s. In the end, the biggest reduction in the spikes can be achieved by increasing bin_s to 20 second. This, however, sounds like a compromise with the temporal resolution so I was thinking if there is alternative method to remove the spikes. One function came to my mind is the motion_cleaner.py, but it looks like I need to replace the high-level correct_motion with low-level APIs and add that function afterwards in my data analysis pipeline. What is your thought about adding an additional argument about motion_cleaner in the high-level correct_motion so that the function can be called directly? Or any thought about how I do motion correction is more than welcomed.

Below is the motion report about my initial settings (the probe has 32 channels and spans around 400 um vertically with horizontal distance 22.5 um)
default settings of DREDge with win_step set to be 75 um and win_scale set to be 150 um

And then here is the result after increasing bin_s to 20s (with win_step set to be 75 um and win_scale set to be 150 um)

Many thanks in advance!