The positive-pair conjecture for SL_n-trace equivalence

Let $F_r$ be a free group, and call two pairs of words $(u,v)$ and $(u',v')$ $\operatorname{SL}_n$-trace equivalent when the corresponding words have equal traces under every representation into $\operatorname{SL}_n$. A pair is positive if, in a reduced representation, its words have only non-negative powers of the generators. Positive-pair conjecture. For every $n\geq 2$, there exist $\operatorname{SL}_n$-trace equivalent pairs $(u,v)$ if and only if there exist positive pairs $(u',v')$ that are $\operatorname{SL}_n$-trace equivalent. The source frames this as a more provocative conjecture after noting that non-conjugate palindromic pairs are expected, but not proved, never to be $\operatorname{SL}_3$-trace equivalent.