On p. 364 of Physics for Scientists and Engineers (9th ed.), Serway and Jewett define a rigid object to be in rotational equilibrium if it has an angular acceleration of zero. They then state that a necessary condition of rotational equilibrium is that the net torque about any axis must be 0, where "any" does mean "every", as confirmed by a later remark in the text.
I am badly misunderstanding something in the above. Consider the example of a rod of length $\ell$ located at time $t=0$ from $(0,0)$ to $(0,\ell)$ in some inertial reference frame. Suppose that two identical and constant forces with magnitudes $F$ act on the ends of the rod, wherever they are, and that these forces are directed in the positive $x$ direction.
I claim that the rod has zero angular acceleration in this reference frame, and is thus in rotational equilibrium. According to S&J, the net torque on the rod about any axis must be $0$. But consider in this reference frame the axis through the origin that is perpendicular to the $xy$-plane. Calculating $\mathbf{r}(t) \times \mathbf{F}$, isn't the torque about this axis always in the negative $z$ direction with constant magnitude $\ell F \not = 0$?
[To clarify: the calculation in the previous paragraph is really a calculation of $$\mathbf{r}_A(t) \times \mathbf{F} + \mathbf{r}_B(t) \times \mathbf{F}$$ where $A$ and $B$ are the ends of the rod.]