Added a couple of TODOs in matrixfill.pyx pointing to possible minus
sign issues.

Added another file, mismatch_mirror.py, to playground/homs/debugging
which shows that Finesse 2 and 3 match for a mismatch at a mirror
when there is no cavity present (i.e. just using gauss commands to
replicate beam parameters from misalign_cavity.py file). This implies
that the issue in that example is to do with computing the round-trip phase
of the cavity (further debugging to do here).