That's an interesting idea. The sad thing is, the money lost through chargebacks is minimal for me in the grand scheme of things but it makes me irate enough to potentially spend the development time (that will never have a positive ROI) on adding in something like this just to stick it to the scammers.
It is at least very simple to implement. It's a simple API. Useful capability for banning abusive users too if you have social elements or other abuse vectors besides payment.