
The saddle-shaped green curve is handle speed against time. The 'catch slip' as he engages the sprocket (on a quite sluggish drive) is visible as the gap between the vertical axis and the left topside of the handle-speed curve. Here it's about 0.01 second as the sprocket engages. If you look at the blue force curve, you'll see a bit of a corresponding 'kink' at the left-hand base of it. That's a sign that he's 'feeling' for the catch with his hands and arms - even though he's actually got the handle going back fast enough to engage the sprocket, he doesn't really begin to lay on the force until roughly 0.04 into the drive. That is, there's a slight hesitation while he waits to feel fully engaged even though he already is sufficiently locked in. But 1) it's in the rower, not the machine; and 2) it involves nothing like 20-30 cm of chain being pulled.
If you propel the handle up the slide faster than the shock cord wants to retract it on the recovery, you can generate corresponding slack in the chain at the catch. This problem is easily solved: slower recoveries and/or adjustment of the shock-cord tension such that the handle retracts more quickly.