Replace linear search in ContinuousTabular with binary search

[realmisch]

Description

Currently, the sampling algorithm in ContinuousTabular::sample relies on two linear searches to locate the corresponding point in the CDF. Since the CDF is monotonically increasing, a binary search can be implemented for $\mathcal{O}(\log\ n)$ complexity instead of the $\mathcal{O}(n)$ complexity of the linear search.

Effect

Performance improvements up to 10% have been found with simple cases and identical results for a fixed seed.
Example case is the Jezebel benchmark from ICSBEP running openmc -s 4 with 500 batches and 100,000 particles. The runtimes are averaged over 5 runs for each search algorithm.

Search Type	Linear	Binary	Difference
Total Runtime (s)	43.139±0.439	39.938±0.938	3.201±1.036
Transport Runtime (s)	27.402±0.280	23.586±0.554	3.816±0.608

Checklist

• I have performed a self-review of my own code
• I have run clang-format (version 18) on any C++ source files (if applicable)
• I have followed the style guidelines for Python source files (if applicable)
• I have made corresponding changes to the documentation (if applicable)
• I have added tests that prove my fix is effective or that my feature works (if applicable)