Turbulence in differentially rotating flows. What can be learned from the Couette-Taylor experiment