I found that these stripes between swaths are probably due to thermal noise, and not to the tops merge operation from this post:
Since the thermal noise removal step also removes slc phase information, I imagine that it cannot be corrected when working with the C2 polarimetric matrix.