Moved filtering into _get_phase_regions function (where it belongs), and have,...

Moved filtering into _get_phase_regions function (where it belongs), and have, for the moment, removed the ConvexHull-based filtering. Also, if alpha is too large, can get disjoint boundaries for a phase region (which isn't inherently an issue) but it can result in ugly overlapping in the plot. So, to avoid this, we make three attempts to remove intersections (leaving us with a solid region) by reducing alpha, and if we fail, we take the disjoint plot. This issue is only really prevelant for systems like LiCl-KCl