I talked to someone a while back (I don't remember if it was here or back in the OP days) that wanted to do this. I told him that I had considered something similar at one time...
Then I went through some of the issues I could remember that I would have to work out before starting. One of which was how are you going to prepare for a servo that goes dead or locks full to one side, while in the air? Quick release? Servo with a spring in the linkage that you can overpower? Trim tab only for control? Full servo can certainly happen if the wrong wire breaks due to vibration fatigue, spilled coffee, rain... For safety, the autopilot must be in a disengaged state any time you are under say 2000 feet AGL (or whatever) or you may not be able to recover quickly enough from a low altitude servo failure.Ahh, I see that you are considering using a trim tab which can be overpowered if there is a failure.
This post assumes you know what you can legally do. I am not going there. And that you will take great care for safety.
LP assisted / autonomous flight is GPS based, so you would have to have a GPS with a good view of the sky, unobstructed by metal, unless you want to write some code. The mags on the FC also need to be usable and that means no strong fields or large ferro metal close by. It would be possible to set what you want up as a ReturnToBase that automatically figures out the direction to fly ... or as a VelocityRoam or PositionRoam flight mode (I don't know the fixed wing details without reading up) that will continue in the same direction you start it in (with no gyro drift issues).
Be aware that our GPS control code does, assumes, requires, automatic levelling. Also be aware that a coded turn is bank and yank (so you really need elevator too), so if you started out pointing the wrong way for whatever reason, you would get full configured bank angle for a long time to get it all the way around and pointed in the correct direction.
With a very stable, nose heavy airplane, and a shallow max bank angle configuration, and starting out pointing in about the right direction, you might reasonably do it with just an aileron trim tab (no elevator), like a single channel RC airplane (but with throttle control for altitude).
There is a "ground vehicle" mode that is "rudder only" if you must set it up maybe on rudder and do the levelling yourself. I wouldn't do it that way.
RTB would require something electronic that acts a bit like an RC transmitter and receiver and generates "centered" stick signals for all the controls you want to have controlled (ail / ele) with the FC set up to fly in manual mode. In manual mode, the servos exactly follow the sticks and the sticks are "centered". This "something electronic" is where you would have a switch to switch from this "manual mode with center sticks" to RTB. A transmitter shaped joystick would work and would be good for testing too. It would also be a good place to trim the trim servos to achieve level flight during testing. You would want to do mechanical trim adjustments to remove the testing trim. This joystick would be good for manual testing of servos / trims to see how well it controls the aircraft.
How are you handling real trim changes if the trims are used for FC control. Make sure that the trims have enough authority to handle actual trim changes plus course correction. Maybe you need a manual trim adjustment as well.
You can't just fall asleep at the wheel or it could gently fly you into the ground or a near by mountain due to reduction of GPS accuracy (e.g. from 3D to 2D). And any set of functions that tries to maintain altitude (RTB does) should be aware of minimum airspeed and know how to handle the case where the motor looses power. With trim tabs only, and a pilot who is always watching ...
I would want to fly a model with similar handling characteristics as the full scale plane to play with settings and tune Rate mode and Attitude mode response times. Your response time will be very very slow on a full scale, and much slower even than that because you are only using trim tabs. What would normally be seen as a oscillation with a frequency of several times per second could be as slow as once a minute or so for you. I would guess this would be OK, but nothing that slow has been tested. You might want to reduce your max gyro rate from 2000 degrees per second to something less to get more accuracy, and that alone would need testing (it's not tested regularly).
Our FC doesn't require an RC receiver, but if I were doing it, I would want at least a transmitter shaped joystick in my lap for air testing manual control of the FC before using the autonomous functions. RC receivers and some joysticks can output signals that the FC can read. Is having an RC receiver connected a legal issue? If not, then an RC receiver that has failsafe would be perfect and no code would be required. You take the transmitter with you for testing and it talks to the receiver and acts like the joystick I described. Once you get testing complete, you set the RC receiver failsafe to have the values that you need for centered servos and fly without a transmitter with centered servos. Disconnecting the receiver from the FC with a switch causes the FC to go into FC failsafe which can be programmed for any mode you want; RTB, VelocityRoam, PositionRoam including the stick positions.
I would definitely use the transmitter shaped joystick or transmitter / receiver to get it working. If you don't want a receiver on board, and you can do some code once you get it working (fairly simple if you already know the code), you can remove the joystick, and hard code the "centered" control sticks and make it think there is a valid joystick / receiver connected when there isn't and use a port (e.g. Flexi or Main) to connect a single switch to force FC failsafe mode.
Have you looked at the code? There is a learning curve.
That is unless you actively want to dive in and understand it. There is a good bit to learn just to make any changes.