The Bezout-based proof is a bit more intuitive interpreted modulo $\,m.\,$ Notice that the Bezout identity $\,\color{#c00}{1 = \gcd(k,m)} = a k\! +\! b m\,$ becomes $ $ mod $\,m\!:\ 1\equiv ak,\,$ hence $\,\color{#c00}{k^{-1}} \equiv a$ exists mod $\,m.\,$ Hence $\, m\mid k\,\ell\,\Rightarrow\,k\,\ell\equiv 0\!\!\!\!\! \overset{\large\quad \times\,\color{#c00}{k^{-1}}}\Rightarrow\!\!\ell\equiv 0\,\Rightarrow\,m\mid \ell.\ $ Therefore, mod $\,m,\,$ the proof amounts to the simple fact that elements $\,k\,$ coprime to the modulus are, by Bezout, invertible, so cancellable.
Remark $\ $ If you know the gcd Distributive Law $\rm\color{#0a0}{(DL)}$ you can write the proof very concisely
$$m\mid k\ell,m\ell\ \Rightarrow\ m\mid (k\ell,m\ell)\overset{\rm\color{#0a0}{(DL)}} = \color{#c00}{(k,m)}\ell = \ell\qquad\qquad$$
The right hand side of the equation is made up of the sum of two terms that are both divisible by m ( (kl)*s and m(lt) ) correct?
– user122661 May 02 '14 at 00:21