It's no wonder you are struggling: What you are trying to prove is simply not true in general.
The quickest way to see this is to consider the special case $n=0$, $m=1$. The integrand $\cos x\over\sqrt{1-x^2}$ is strictly positive for $-1\lt x\lt1$, hence
$$\int_{-1}^1{\cos x\over\sqrt{1-x^2}}dx\gt0$$
If you dismiss this counterexample by requiring $n$ and $m$ to both be positive, then consider $n=1$ and $m=2$. Wolfram Alpha finds
$$\int_{-1}^1{\cos x\cos2x\over\sqrt{1-x^2}}dx\approx0.793481$$
(and an exact value in terms of Bessel functions). It shouldn't be too difficult to give a non-Wolfram proof that this integral is positive, but I don't see any easy way to make it obvious.
It seems likely that the integral in question is never equal to $0$. Wolfram Alpha finds in general that
$$\int_{-1}^1{\cos nx\cos mx\over\sqrt{1-x^2}}dx={1\over2}\pi(J_0(m-n)+J_0(m+n))$$
where $J_0$ is a Bessel function of the first kind. A hand-wavy reason why you don't get $0$ out of this is that, while the limits of integration here, $\pm1$, are "natural" values for the square root function in the denominator, they are not particularly natural for the cosine functions in the numerator, for which rational multiples of $\pi$ are more natural. There might be an actual proof based on properties of Bessel functions; I'd love to see one.