Hm, Loop Hair. While possible it's not trivial as far as
I can see.
Thanks to no cache offset it looks as if PLA baking is the option at hand.
In rough terms, use a MoSpline with the Hair Object as source.
Then run a script to make a editable version of the MoSpline and
read the PLA values for each frame while baking it.
And then do the PLA looping described earlier.
An addition option is then to re-bake the sweepNurb to a
PLA baked polygon object, run a PointCache on that.
This way the animation can be slowed down -much- more
thanks to the PointCache's higher subframing.
A couple of first tests (loops are perfect but playback with H264 is not). HairToPLA 9mb QT HairToPLA02 7mb QT