We had previously agreed that demonstrating that the motion of the mirror has a component perpendicular to the surface of the mirror was sufficient to prove that doppler shifting would be observed in light reflected by the mirror.
Therein lies your repeated error.IF you insist in transforming the velocity of the mirror into the "camera frame" (for reasons I cannot fathom), THEN you also need to transform the normal to the surface. Only AFTER that you may compute the angle between the velocity and the normal. You are transforming only the velocity and you forgot to transform the normal in rushing to declare that the mirror velocity has a component along the facet normal as viewed from your "camera frame". Turns out that it doesn't.
Since you like Galilean relativity, since it is simpler, I will use an explanation that works in Euclidian space, for subrelativistic speeds. You have often resorted to this simplification, so I hope that you will not object. In Euclidian space, angles are frame invariant, I will follow a proof that you can also find in Rindler (Relativity, Special, General and Cosmological, second edition, page 96).
Let's consider two arbitrary vectors, $$\vec{a}$$ and $$\vec{b}$$. Their scalar product is an invariant. Indeed:
$$||\vec{a}+\vec{b}||^2=a^2+b^2+2 \vec{a} \vec{b}$$
But, in Euclidian geometry, the norm is an invariant, so:
$$2 \vec{a} \vec{b}=||\vec{a}+\vec{b}||^2-a^2-b^2$$
is an invariant, where $$a=|| \vec{a}||$$ . QED.
But there is more, the angle of the two vectors is also an invariant since:
$$cos(\theta)=\frac{\vec{a} \vec{b}}{ab}$$
Now, what I am asking, is whether or not this is equivalent to demonstrating that the camera has a component of motion perpendicular to the surface of the mirror.
Not, it doesn't. See above, angles are frame invariant. A 90 degree angle transforms into a 90 degree angle.