Triangulation: Add static filter for 5D and 6D orientation predicate

[afabri]

Summary of Changes

This PR adds a static filter for the orientation test of 5D and 6D points. It is ~20% faster than just interval arithmetic.

Release Management

• Affected package(s): Triangulation
• License and copyright ownership: unchanged

[mglisse]

So far it is slower than just interval arithmetic

I tried quickly running the benchmark (on a file generated with the new generator).
With master, I get 11s.
Branch with -DCGAL_NO_STATIC_FILTER_5 (so presumably just the clear/mark change): 10.3s
Branch (with static filter): 8.5s

That doesn't look slower to me 😕

• The TDS change is exactly what I had in mind
• orientationC5.h doesn't seem to be used (ah! it was probably used as input to FPG, then it makes sense to keep it somewhere indeed, with a comment to avoid confusing people like me 😉)
• we could write the static filter directly with the internal interface instead of defining an adapter (I used the adapter approach for 2d and 3d because I wanted to reuse the static filters from Kernel_23 without modifying them), but if you prefer with the intermediate adapter, that's fine with me.