I would agree that thrust is actually created by the inlets. I would also agree that it is counter-intuitive. I had to write out the momentum equation to convince myself that it was actually true. Odd as it may sound, during diffusion in the inlet, the fluid exerts a force in the forward direction on the inlet, or a thrust force. Think about an axial compressor as an analogy. It compresses the air (obviously) but it also produces thrust. It's a simple application of Newton's Action-Reaction Law, you have lots of little airfoils pushing the flow backward, so the flow is in turn going to push those airfoils forward. Now for the SR
-71 I guess I'm going to have to take people's word for it about what percentage of thrust is produced by the inlets.
In normal supersonic operation, the inlet cone is positioned so that the oblique shock coming off of it enters the inlet. It then proceeds to bounce off the inlet walls until, compressing and slowing down the flow. By having the oblique shock bounce off the walls, the flow must go through a series of them, each one compressing and slowing the down a little bit. Eventually after the flow passes through all the oblique shocks, it has a higher pressure and is traveling just over sonic speed. Before it heads off to any other part of the engine, the flow goes through a weak normal shock to raise the pressure a bit more and bring the speed down to subsonic. This subsonic compressed flow can either enter the compressor or continue around the engine core and go straight to the afterburner/ramjet burner. (Reducing the flow to subsonic speed is very important since the Blackbird's ramjet is not a scramjet.) Note that some of the air (albeit a small percentage at high speed) always goes through the engine core; the Blackbird does not operate as a pure ramjet at any time.
An "unstart" occurs when the pressure gradient across the inlet changes due to some disturbance. The oblique shock then no longer enters the inlet, which means the engine loses all the compression/thrust that the oblique shocks in the inlet contributed. As was mentioned, Ben Rich and Co. devised the "sympathic unstart" in the unaffected engine to alleviate the problem.