Bayes benchmarks

[turion]

• [x] Merge https://github.com/turion/rhine/pull/221 first
• [ ] Clean up Git history
• [ ] Create a test suite that runs the oscillator example with a simulated clock
• [ ] Measure inference quality in terms of mean & variance
• [ ] Test that for fixed RNG, these quality indicators do not change significantly from the current value, over several different runs
• [ ] Perhaps using https://github.com/Bodigrim/tasty-bench-fit, ensure that complexity is constant over time