I don't think #2 is really even possible with hobby rockets. I've been thinking about augmenting a rocket with active roll control using a cold gas thruster. The simple fact of the matter is, though, I don't think any RSO would let a rocket fly that is not already passively stable. On top of that, you would have to carry around pressurized tanks and other heavy equipment. A cool idea, though.
Oh, gimbaled motors have been done. John Pursley (former editor of American Spacemodeling magazine, before it became "Sport Rocketry") has done it. I've seen his work firsthand, and it's something to behold.
Basically he adapted a regular, off-the-shelf model airplane anti-crash autopilot system for use in a rocket, with a sensor ring with four sensors arranged like the cardinal points of the compass... these look out toward the horizon from just under the nosecone of the rocket. These sensors detect the horizon line by detecting light at a wavelength opaque to CO2, therefore the atmosphere above the horizon (sky) appears "black" to the sensor, while the much shorter distances to the horizon and land reaching out to the horizon appears "white" to the sensor. By comparing opposing pairs of sensors outputs, which ideally should both be reading 50% sky, 50% "land" below the horizon, the flight computer can guage the degree of tilt of the model its installed in (plane or rocket). If one sensor is seeing "all white" (looking down toward the ground) and the other is seeing "all black" looking upward into the sky) then the computer knows the vehicle is tilted in that assigned direction (be it pitch or yaw). The other pair of sensors 90 degrees to the first pair work similarly, so the two opposing pairs of sensors detect both pitch and yaw. There is no detection of the roll axis nor correction for it, but it isn't particularly necessary in a finless rocket anyway.
Once the flight computer (anti-crash system) detects the orientation and tilt of the rocket, it generates a corresponding and proportional servo signal to correct that orientation. SO, all one has to do is build a homemade motor mount with a swiveling gimbal (U-joint, basically) with two servos 90 degrees apart from each other-- one to swivel the motor mount back and forth for pitch control, and the other to swing the motor mount side to side for yaw control. The two can work in tandem to swing the motor in both directions at the same time, according to inputs from the stability system. The main thing is to ENSURE that the proper servo is activated by the proper pair of sensors.
Now, the OP is correct in that basically at burnout the stability contribution from the gimbal system is greatly degraded... Even a real rocket that has an engine shut down has no control authority from gimbaling that engine. Thing is, our solid rocket motors DO produce SOME VERY SLIGHT "thrust" from the gases produced from the delay grain (especially in BP motors) and this does contribute SOME control authority during coast... the servos may have to "hard over" (extend to full control throw) to correct the flight path, but from John's experience, the system DOES work, even on a finless rocket. Of course if the rocket has some kind of small, undersize, or SCALE size fins, this helps enormously. John has flown completely finless rockets with this method and had it work beautifully... His Vanguard rocket flight was amazing to behold! He's also used the system in a scale Mercury Redstone with scale-size fins with excellent results, though of course the Redstoned DOES have some smallish fins on it. There's really no reason with proper ballasting that a Gemini launch vehicle could not be similarly flown with such a system. (Ballasted to move the CG ahead of the CP through the use of noseweight).
That said, the rocket needs a certain amount of "natural stability" to make the task easier and for the gimbaled motor to continue to work after motor burnout when its control authority drops precipitously. The Vanguard rocket was an ideal shape for this sort of setup-- the long thin upper stages, with the wider first stage, meant that the CP was further back than on a lot of designs. This setup is NOT well-suited for a "hammerhead" rocket design like say, a finless Ares I, since its design is COMPLETELY THE OPPOSITE, with a large, long, fat upper stage above a long, skinny, much thinner first stage, which will shift the CP forward, requiring an inordinate amount of noseweight to move the CG ahead of the CP or at least to a neutral condition. If the rocket isn't ballasted right, at burnout it will most likely "swap ends" and do a backflip and try to fly backwards through the coast phase...
The important thing is, IT CAN AND HAS BEEN DONE. Electronically controlled roll control has also been done, using a servo to move a small pair of fins in opposing directions to null out any roll rates for a camera rocket. The servos were controlled by a ring of photocells located around the rocket body tube in a half-circle (IIRC) configuration. Each photocell was wired in such a way to detect sunlight falling upon it and convert that into a voltage reading. The electronics were hard-wired in such a way to produce a servo input signal to control the servo and proportionally move the fins depending on which photocell the sunlight was falling directly upon (therefore generating the most voltage) and steer the fins to turn the rocket until the "center" photocell was directly facing the sun-- if it rolled past, the photocells on the other side of it would detect more sunlight, and reverse the servo proportionally to turn the rocket back so the center photocell was facing directly into the sun. The system worked beautifully. It's on the Brittain Fraley website. I would also note here that George Gassaway's "Sun Seeker" rocket used a similar system of a ring of photocells INSIDE the rocket, with sunlight streaming in through a window in the tip of the nosecone (IIRC) to "steer" the rocket so it flew directly toward the sun after liftoff, via servo-controlling the fins. Such a system could also control a gimbaled motor on a finless rocket if one were so inclined to develop it that way.
Later! OL JR