Implement scipy detrend with 'soft_l1'

Question

I needed to use scipy detrend, but my data display various peaks, some legit some not. I removed most of the unwanted ones by hand but the legit ones are creating some probl