It is pretty amazing how we can look at the same clip and see completely different things.
FOD,
If you look closely enough (it helps to have the entire clip) Lincecum really hasn't started opening his hips at all as of frame 88. His stride leg has opened up but not his hips. Most people can't do that and they start opening up earlier as a result. Then his hips, and his torso/shoulders open up very quickly and pretty much together although the hips probably lead a bit. Given that Lincecum counter rotates he may actually begin the shoulders rotating about the spine prior to the final hip rotation. The hips and torso rotate both due to muscle action and due to momentum transfer from the plant leg slowing the forward momentum. His shoulder externally rotates in response to the hip and torso/shoulders turning. It reaches full external rotation and as he squares up the shoulder rotation about the spine virtually stops allowing a combination of the "stretch" in his arm and the momentum transfer from his body to the shoulder catapult the ball forward through internal rotation of the shoulder and to a lesser degree extension of the elbow. There's a bit of forward motion of the shoulders/torso that contributes a small amount at this point. He gets fully extended at release without pronating at that point. As the ball is released and the load from the ball goes away the forearm pronates to keep from further loading the elbow. The damage to the UCL tends to happen during acceleration prior to release and the damage to the bony structure tends to happen just after release if there isn't pronation just after release although the bones have begun to offload the UCL at near full extension prior to release. After release the shoulders rotate to relieve the load on the shoulder during the follow through. Even though the elbow is extending like crazy prior to release it doesn't hyperextend until the ball is released abruptly relieving the load of the ball acting against extension so pronation immediately after release before the elbow can fully hyperextend is vital. From looking at that clip it is possible that Lincecum is getting fully hyperextended and that the post release pronation is minimizing the effects by not allowing the elbow to remain hyperextended through the entire follow through. Hard to tell.
Any pronation prior to release will tend to have a significant negative effect on velocity due to lack of extension and not having the fingers directly behind the ball. If one is willing to give up that velocity then it should be a bit safer and possibly result in more movement and less ability to locate. Screwballs can be very effective changeups and a lot of pitchers pronate a bit early on their changeups to get more movement and to take off a bit more velocity.
I can see where the Marshall technique could result in max velocities almost up to what a typical cutter would result in and if a pitcher could repeat their motion consistently enough then they might be able to be fairly effective and a bit safer with the motion. It doesn't allow for the same max velocity from a given pitcher as conventional mechanics though.