Failure to properly initialize early values of the super smoother will result in misleading values early in the output. Because the SS is an IIR ( infinite impulse response) filter, this error can ring in the filter for a long time, but is extremely evident in the first 2*len bars.
This is an implementation if the 2 and 3 pole SS filter, with special attention to initializing the early values.
It uses (src+scr)/2 per Ehlers but contains code to just use src if you prefer to calculate that outside the function as everget does in his SS here: there is code included to make that change.
Many thanks to everget for his terrific implementations of much of John Ehlers' work. It has been tremendously helpful to me.