What I thought we needed to do: Create a docker image, bring an ARM Qemu image up, and build it there.
What I found it would be easier: Just use the qemu user mode, which runs a single command in another architecture. So instead of calling cmake inside a Qemu ARM image, I can qemu-aarch64 cmake.
What is the new issue: For that to happen, I need to have all the ARM building tools available. One solution would be to have a volume with all the ARM executable to build something, mount it along the image with qemu-aarch64, so when it calls gcc, it would actually call the ARM version of gcc.
And I'm not sure how to implement that last point.
