POCA should use smoothed phase difference

[j-haacker]

While not important, I believe it would be cleaner to use the smoothed phase difference for the POCA sample.

I believe so because if viewing the footprint as slice in across-track and elevation, the transmitted signal only touches the surface at the POCA (point interesection). The left and right intersections of consecutive samples are close. In 3d, the problem is similar, but we expect that the samples all. even before the POCA, interect the surface at higher elevations ahead of or behind the nadir point in along-track direction. Statistically, I expect the echos contributing to the samples around the POCA sample to originate from about the same across track position, such that averaging the phase difference of a few samples rather reduces the statistical uncertainty than introducing a bias. While we could think of topographies where, contrary to the regular case, a bias would be introduced, I expect those cases to be very rare and the error to be small.