Following the discussions in issue #177 and forum post #221 we need a dedicated function
for removing polarization vectors and spinors in the given amplitude.
It would be also good to have the resulting open indices somehow related to the ordering
of the original particles (so that one can use projectors later on).