refactor: abstract LocalSearch stats into LocalSearchStats

refactor: Use OVector instead of SVector in library

feat: add evolutionary strategies to local search

fix: put Rng to Box instead of generics

feat: add test for local search with PatternPerturbation

feat: add PatternPerturbation

Move by +/- d in random single coordinate.

feat: add uniform constructor for RandomDistributionPerturbation

feat: remove min, max from bounded perturbation

tests: add test for Linear fitness function

feat: add BoundedPerturbation

applies bounds on real numbers when performing the perturbation.

feat: add RandomDistributionPerturbation

For vectors of reals, perturb by a random distribution with
expectation in 0.

feat: add real fitness functions

Add linear, step, rastrigin, griewank and schwefel fitness functions
along with the tests for them.

Try using static vector where appropriate, but use OVector for those
where filling with zeros won't work. This is mostly a constrain for
the tests... rather than a real constrain. In the end the dimensions
will be given for each problem.

refactor: extend local search to accept any inputs

Instead of accepting only BinaryString, accept anything.
Extend the operators to accept anything.

tests: move all tests to separate submodules

tests: add rosenbrock test

feat: add MaximizingOperator

chore: guard test imports by #[cfg(test)]

chore: split types and functions to separate module files

tests: let EqualTerminatingCondition remember matches to return even if no longer equal

tests: For local search test, wait 100 cycles after optimum found