When dealing large models, the Moose Finder was slow. I looked into
the reason, and it was due to Group>>= which was doing a
transformation to OrderedCollection.
I now removed the method, and thus group equality relies on the
default == implementation. All tests work, so I guess it is not really
Anyway, as a consequence, the Finder is significantly faster now.
Please let me know if you relied on Group>>= or if you encounter other
"Every thing should have the right to be different."