Is there some knot theory behind the Mobius donut? I was watching this video by Numberphile where a professor cuts a bagel into two interlocking pieces. Is this a torus knot or torus link?
I'm trying to interpret in terms of $(p,q)$-torus knots Torus knot (Wikipedia), but it seems like the cut goes once around the axis of rotational symmetry, and once around the circle in the torus, but I think I'm interpreting it wrong because since $\gcd(1,1)=1$, you shouldn't expect two components, even though that is what happens. 
 A: You might be familiar with how a torus can be thought of as identifying sides of a square.  If not, look at this little gif from wikipedia: Identification square. A torus knot can be though of as a line with rational slope that passes through the square and when you get to an edge, you move to the opposite edge and continue on at the same slope. 
So, a $(1,1) Torus knot is the diagonal of the square, which looks like this.

But if you were to take a cylinder and stick a knife through it, you would cut it in two places.

If you were to follow the line as it cuts through the torus, or for us, the cylinder (we are cutting it before it would be bend around to make the usual torus) , you will see we get two independent  lines that intersect the cylinder.  One that starts at the bottom and one that starts at the top, which the second picture shows in red and black, respectively. 


So, the corresponding identification square should have two $(1,1)$ torus knots, one in black and one in red. 

Hope this was helpful.
A: Every half donut is orientable, since it has two faces, hence it has two border components, and, as you suggest, they are both $(1,1)$ torus knots. The problem is that they are two torus knots, not one, hence you can get two connected components.
