I support CyclesMedernach with the following changes:
Remove the words "newly allocated". Since cycles are immutable, there should be no way to tell if they share structure or not.
Clarify in cycle-rotate that k can be positive to rotate forwards (relative to the list passed to make-cycle) and negative to rotate backwards.
Add cycle, cycle-ref, cycle-for-each, and cycle-fold by analogy with R5RS list, list-ref, for-each, and SRFI 1 list-fold. (I believe that in WG2 we should always have a fold procedure for container data structures.)
Add cycle=? which checks that the contents of the cycles are the same (in the sense of eqv?) without regard to the current position of the cycle.