Resolve "(core): parallelize linter execution of glslang_validator when handling multiple macro combinations"

Description

This parallelizes the glslang_validator hook execution, to speed-up the lint with shaders containing many macro combinations.

Closes #14 (closed)

How to test it?

• Modify a glsl and add several #ifdef/#endif couples (eight for instance)
• Stage the file and run sheldon
• Observe the execution is much faster with the MR than without