I've cherrypicked the appropriate branch of arcosh to allow the method to 'work'.
Is this a valid integration method?
Using the substitution $x=\cosh\theta$ we get:
$$\int_{-1}^{1}\frac{dx}{\sqrt{1-x^{2}}}=\int_{\operatorname{arcosh}\left(-1\right)}^{\cosh^{-1}\left(1\right)}\frac{d\theta}{i}=-i\left(\cosh^{-1}\left(1\right)-\operatorname{arcosh}\left(-1\right)\right)=-i\left(0-\left(-i\pi\right)\right)=\pi$$
I would concede that it's rather convenient to simply choose $\operatorname{arcosh}\left(-1\right)=-i\pi$ so I'm wondering if in fact it's even a legal move.
In solving $$\frac{e^{w}+e^{-w}}{2}=-1$$ it can be observed that $w=i(2n+1)\pi$ where $n∈ℤ$, so I'm left wondering what exactly justifies choosing $$\operatorname{arcosh}\left(-1\right)=-i\pi$$ instead of choosing any of the infinitely many other options, such as $+i\pi$