Part Holds ignore boundaries after placement

[Thomas Berezansky]

It appears that the hold targeter ignores selection depth/ou for Part holds. This has the effect of boundaries not being obeyed for part holds once placed, provided the hold was able to be placed in the first place (as the hold placement tests seem to be fine).

I believe the solution is probably along the lines of a new ranged_tree function for parts, and then teaching the hold targeter to call it.