I am looking for the origin of the following idea: suppose $m$ and $n$ are relatively prime integers $\geq 3$. Let $E$ be an elliptic curve over a number field $K$. Let $L/K$ be a finite extension such that the torsion subgroups $E[m]$ and $E[n]$ are contained in $E(L)$. Then $E_L$ has everywhere semistable reduction.
This follows easily from Raynaud's criterion [SGA 7 I, Exp. IX, Proposition 4.7], but I cannot pin down the first appearance of the above version of semistability for elliptic curves.