Recall earlier we were talking about the delay between what the pilot sees (because of the delay in that data being transmitted from the aircraft to the control facility), and the additional delay for his control inputs to make it through cyberspace to the aircraft. The length of that two-way delay is such that the ability to maneuver real time tactically is just not feasible. The Reaper is not armed with a gun, but even if it were, the maneuvering limitations on the airframe, combined with the delay in maneuver inputs from the pilot, make strafing just not practical. Add to it the fact that strafing is used specifically because it is a discriminatory weapon: gun effects can be focused on a small area on the ground, and thus is useful for situations in which friendly and enemy forces are physically close together (too close together for the weapons effects of a gravity bomb). The current control methodology for RPAs is just not precise enough, nor is the aircraft performance precise enough, to allow precise aiming and shooting of a gun in the vicinity of friendly forces.
As I've said before on JC, it is inevitable that there will be UCAVs in the future: remote operation is a methodology of flying an aircraft, and just because the aircraft that are being remotely operated NOW don't have the performance required to operate in a hostile/denied environment, it doesn't mean that future iron won't. We will definitely have those aircraft in the mid-future (15 years, maybe?), and I'm sure the control technology and bandwidth will increase to allow more rapid and precise control of those aircraft. Will one of those aircraft be a CAS aircraft? In my opinion, CAS will be one of the last missions to lose actual piloted aircraft. There's all ready been some backlash against RPA use in CAS which resulted in the U-28 and MC-12 being designed and built specifically to have similar capabilities to the RPAs but with human eyeballs on board able to look out the window.