Finding the correct correlation window size

[scottstanie]

There's an open TODO in the code now:

https://github.com/isce-framework/dolphin/blob/main/src/dolphin/workflows/displacement.py#L130-L147

• Before I had the window size come from the phase linking window, like:

row_looks, col_looks = cfg.phase_linking.half_window.to_looks()

It's been changed to be square since it was not accounting for the strides: we used by default a (11, 23) multilook search window for SHPs, but after strides the pixels are square. We were not adjusting the window for the correlation estimation.