This task will allow the user to compile all kernels that are used in foreground rendering.
In the preferences there will be an operation to compile the kernels. this operation will only be available when OpenCL is selected.
with recent changes there are 32 variations of the kernels + one preview kernel. Compiling all variations will take around 45 minutes on a Ryzen 1700 machine.
It uses around 20 GB of memory as it loads the kernels in the cache.
We should try to not load the kernel in the cache
When not loading in the cache compiling takes 0:30:59.775920. I see that the task pool with external processes misses some compilations and use the fallback.
Perhaps better to use a poll mechanism so we don't need to restart to many tasks