oh now I think I know why you titled this as a statically indeterminate problem. Your first example appears to be a statically determinate shaft fixed at one end and free at the other end with a Torque applied at the free end. For a statically determinate problem, the variation in cross sections doesn't matter when computing torques at various sections of the shaft. You just use the equilibrium equations and free body diagrams to solve for the twisting moments at any section. If for this statically determinate shaft you were to apply a clockwise torque T at some point between the ends of the shaft, then the torques left of that point at any section toward the fixed end would be T counterclockwise , and there would be no torques in the shaft right of that point toward the free end . This is strictly from equilibrium considerations using sum of torques = 0.
Now in your last example with the shaft fixed at both ends, you have a statically INdeterminate problem, which is whole new ballgame, because internal and end reaction torques are very much dependent on the various section parameters (J, G, and L). So in general T1 and T2 are not equal, and you need to look at twist deformation angles to solve the problem. Here both end reaction moments have the same direction as you note, and the angle of twist at the change in section shapes is the same on each section at that point, one considered positive and the other negative, by convention.