Monte Carlo integration is a technique frequently used in offline rendering; however, it is a very versatile method that also finds much use outside computer graphics.
I posted an example question earlier today: How do correlated samples influence the behavior of a Monte Carlo renderer?
These kinds of questions seem related to computer graphics, although you could argue that they are better suited for statistics- or math SEs.
Do we want to allow questions about Monte Carlo integration?
To me the answer is clearly yes. Monte Carlo integration is, like you said, the underpinning of a huge area of computer graphics but, more vitally, this area (path-tracing and offline rendering) is the one most under-represented in StackExchange right now: Real-time rendering is mostly considered on-topic in the GameDev SE, and if you restrict general math that finds a common use in CG, you'll end up removing a lot of the point of this SE existing, which is to house more theoretically-minded questions, which naturally will involve more theory and math.
I think there's no possible better outcome than real-time graphics folks sharing this site with film and animation graphics folks. Monte Carlo, though it is a statistical technique, is so fundamental to offline rendering that it cannot possibly be excluded. It would be like exiling every question that mentions radiometry or photometry to the Physics StackExchange.
