#11 Dominances out of roots

[Alexander Koller]

I'm not sure that the solver currently solves
constraints correctly in which both the root and a hole
of the same fragment dominate nodes in the same hn
connected component. In such a case, the sub-solved
form of the hncc should of course go below the hole --
but the current solver might also record it below the
"other" children. We need to check this and if the bug
is there, fix it asap.