Well, if enough air gets trapped in the tubing, the flow can actually slow down. I don't know how much air this would take, but in small bore tubing systems (such as in an intravenous drip) a few inch buble can really slow the flow. Exrapolating to a python size tube, I would think it would take several feet of air.
As for when you first start up your python, it seems to me to be slower when there is not water in it at the start, then speeds up as more and more air is eliminated. You have to be careful that coils in the tubing don't give the optical illusion that that things are speeding up in the coils compared to the straight parts. I have never really studied or tested this though.