Try new spatial matching methods

[bcipolli]

From @atsuch:

i also want to try a different way of RL matching. Rather than find a spatial match by directly comparing R and L components as currently implemented, do R- and L- matching through wb components

[atsuch]

I implemented it in the newRLmatch branch, but I found that in some cases signs of the maps are causing an issue. I think it would have been a problem regardless of the matching method, but basically, the scoring method doesn't care if the two maps have opposite signs, and it could match R and L maps with opposite signs. But that would create a concat RL image with flipped signs on R and L hemisphere, and such image becomes hard to match with any wb components.

Example here.

[atsuch]

To fix it...maybe we can modify compare_components in decomposition.py so that it flips the sign of the second image if it produces the best dissimilarity score and return the image pair? What do you think, @bcipolli ?

[atsuch]

The main.py in newRLmatch allows you to choose between the original R to L direct matching and new RL matching through wb.