I have my doubts that enough energy could be captured from the minute movements of pointing and clicking to power a wireless mouse. An optical mouse is out of the question, as that works by bouncing laser light off of the desktop continuously and looking for changes. That cannot be paused, or some motion will not be captured, and that makes it energy-intensive.
Thus this can only be done with the old ball style of mouse, with the spinning of the ball both powering the mouse and signalling motion. There a trade-off has to be established between signal sensitivity and power capture, as to capture more power you have to remove more of the energy from the ball, leaving less to be measured for signal. It's a nice idea in principle, but the implementation seems more trouble than it's worth.