breaking: Unified Random Number Generator

[KSkwarczynski]

Pull request description

To be able to reproduce MaCh3 chains introduced new way of accessing random number by singleton.
This mean it is much easier to reproduce problematic fit behaviour.

I sticked to TRandom based on tests, but we can in future quite easily switch to STL.

Yes, new approach is fully thread safe.

Credits to David and Luke for inspiration and suggestions

Examples

4 chains with same seed (10) run independently produces same posteriors.

I tried benchmarking new approach so it isn't slower than old one:

[Benchmark.txt](https://github.com/user-attachments/files/31147560/Benchmark.txt)

---

• I have read and followed the contributing guidelines.

[github-actions]

Hi @KSkwarczynski, thank you for contributing to MaCh3! Please wait for MaCh3 developers to review your PR. If no one answers within a week, please message people from this list: https://github.com/orgs/mach3-software/teams/mach3admin . While waiting, please enjoy this

Use this action on your projects. Use jokes on issues instead.

[github-actions]

Super-linter summary

Language	Validation result

All files and directories linted successfully

For more information, see the GitHub Actions workflow run

Powered by Super-linter

[github-actions]

Hi, I am MaCh3-Telemetry bot

Memory and CPU Usage Summary

### Memory and CPU Usage Summary

| Metric  |     Mean    |      Max    |
|---------|-------------|-------------|
| **RAM** |   357.23 MB |   365.09 MB |
| **CPU** |    96.38 %  |    97.75 %  |

Memory Usage Plot

• Job Summary Link

You can view the memory usage plot directly in the job summary from the link above.