Scenario 1:
Here, $P>Q$. $O$ is the center of mass of the rigid and uniform bar/stick. The resultant acts to the right of $\vec{P}$ as $P>Q$.
Scenario 2:
Here, $P>Q$ also. $O$ is the center of mass of the rigid and uniform bar/stick. Now, the problem here is that as $\vec{P}$ and $\vec{Q}$ act at the two ends of the bar/stick, there is no place left to the right of $\vec{P}$. So, where will the resultant of magnitude $(P-Q)$ act?



It's more useful and natural, and less confusing/misleading, to label vectors with their names $(\mathbf Q,\mathbf P,\mathbf P+\mathbf Q)$ rather than their lengths $(|\mathbf Q|,|\mathbf P|,|\mathbf P|-|\mathbf Q|).$
Likewise, lengths $(|\mathbf Q|,|\mathbf P|)$ ought to be notationally distinguished from points $(Q,P,O).$
Assume that the rotation axis passes through $O,$ that all forces are vertical, and that $\mathbf Q\ne\mathbf0.$ Then the resultant force-torque $(\mathbf P+\mathbf Q),$ which has the combined effect of $\mathbf P$ and $\mathbf Q,$ indeed has a vertical line of action on $\mathbf P$'s right. Obviously, this resultant force cannot be physically implemented.
(Circling back to my opening point: “resultant of magnitude $(P−Q)$” is confusing to parse, unlike “resultant force with magnitude $(P-Q)$” or simply “resultant force $\mathbf P+\mathbf Q$”. Still, since we're already using unboldfaced capital letters to denote points, $|\mathbf P|$ is better than $P$ to denote $\mathbf P$'s length.)