Add intermediate course

[pzehner]

This course should cover the following sections:

• Profiling and debugging tools;
• Subviews;
• Atomic operations;
• Hierarchical parallelism (tough one);
• Atomics;
• Scatter views? (exp);

Lower priority:

• Effect of layout;
• Kernels:
  • How kernels are launched;
  • Functors;
  • Another point I forgot.

Idea to move the "Extending loop policies" and "Parallel reductions" in the intermediate course and move "Profiling and debugging tools" in the basic course.