Physically-based Sky Model based on Nishita93

[joelbarmettlerUZH]

This PR implements a new environmentEDF based on the Nishita93 sky model. The sky model was implemented by me as part of a university thesis with support from @LZaw.

The here proposed sky model, referred to as "Applesky20" within my thesis, implements Nishita93 single scattering sky model but extends on Nishita93 by adding the support for Ozone. A comparison of the here proposed "Applesky20" model shows that it can easily compete with the new physical sky model released by Blender v2.92 a few months ago. A port of Blender Cycles sky model can be found in my fork of appleseed.

Full implementation details are given in my bachelor thesis about phsyically based sky models in appleseed.

This PR is marked as "WIP" since the following issues must be resolved first:

• [ ] Sky brightness issues must be discussed in the appleseed #dev-physical-sky channel. Fore more details, please read the first paraph of Subsection 6.3.1: Visual Comparison of my thesis. I will open up a new discussion thread in discord.
• [x] There are some minor compiling issues on Linux that must be resolved first, mainly regarding the CMAKE file. These issues can be resolved rather quickly and they only affect Linux users (to my knowledge).

None the less, a first review of this PR would be much appreciated, primarily focusing (at least) on the following:

• Compliance with appleseeds coding standards
• General structure of the classes and files
• Correctness of the overall algorithm
• Inefficient, overcomplicated or suboptimal functions or routines

Note that a full review (testing with demo-scenes etc.) will only be possible once the three points above are resolved.

[joelbarmettlerUZH]

Thanks for the review! I'll try to resolve most of the commends next week, maybe we can schedule a meeting afterwards