First of all, the optical assemblies were mostly about determining attitude. Position was determined onboard by dead reckoning from periodic state vector updates from the ground. The ground can obtain a very accurate position by observing the direction of the radio signal over precise intervals, whether in Earth orbit or in the translunar orbit. The optics were rigidly aligned to the guidance reference platform by some extremely robust structure. Hence the optics could tell you if the platform had drifted by even a small fraction of a degree. With some math and careful observation, it would have been possible for the the crew to use the sextant to observe landmarks on Earth and reason about their position in orbit, but that was a contingency technique.
The extent to which the crew had manual control was simply the extent to which they wanted it. They could configure the flight controls to give them manual control over the SPS and the RCS, the latter being capable both of rotations and low-impulse translation. The system was fly-by-wire, but the control system could be configured to apply the hand-controller inputs in different ways. For example, in one mode you could deflect the hand controller and the ship would rotate at a speed proportional to the amount of deflection until you released the controller, whereupon the ship would stop. In another mode the rotational acceleration would be proportional to the degree of deflection, so that you could do things like set up a constant rotation rate. Finally, in "hard over" mode, the hand controller would fire the appropriate jets directly -- bypassing the fly-by-wire -- if the controller is deflected all the way in any direction.
For ordinary attitude control it was simply easier to punch the pre-planned attitude values into the digital autopilot and let it do its thing. But nothing prevented full manual control.
