Unification with HIP

[robertodr]

This is an attempt at extending this lesson to include HIP examples alongside the CUDA ones. This is based on discussions in the Zulip chat.

• Change name of lesson to "Heterogeneous computing with CUDA/HIP"
• Add an episode on performance portability and HIPification tools. This will be based on the HIP101 materials put together by CSC
• Change code examples to use sphinx-lesson tabs, such that one can toggle between the CUDA and HIP versions.
• Update code samples and exercises to have a HIP counterpart. This can be done with hipify-clang or hipify-perl.
• Extend custom.py to also point to the HIP documentation for functions. Might not be quite smooth, as the HIP online documentation is not as well-organized as for CUDA.