Move new samplers out of experimental module and put old samplers at legacy module

[chaozg]

General description of task

Move the "new" samplers at cuqi.experimental.mcmc to cuqi.sampler and the "old" samplers at cuqi.sampler to cuqi.legacy.sampler.

Definition of Done (Feature/change)

• Move samplers at sampler, including all subsamplers and gibbs, to legacy.sampler.
• Move samplers at experimental.mcmc to new module 'sampler' and keep it under cuqi.
• Update all demos under dev, howto, and tutorial
• Update unit tests
• Update the logic in BayesianProblem and use the new samplers by default with BayesianProblem
  • Note: From now on, the default samplers of BayesianProblem are changed, from old/legacy to new.
  • We can still use old/legacy samplers with BayesianProblem by providing a flag legacy=True
• Fix problems in unit tests related to the above change to BayesianProblem
• In the main legacy.sampler.Sampler class, we add a warning once the user creates an instance of the old samplers (including legacy.sampler.Gibbs, which doesn't inherit from legacy.sampler.Sampler)
• Rename test file test_sampler.py -> test_legacy_sampler.py
• _recommender.py is moved to the same level as the new sampler files, i.e., sampler/_recommender.py. It is decided to leave it at experimental after group discussion.

Definition of Done (Mandatory)

• Documentation added (docstrings on all public methods/classes)
• Unit tests added/updated (and passing!)
• Code reviewed and approved by 2 CUQI-maintainers
• Online documentation is rendered correctly (esp. math)

[chaozg]

Hi @jakobsj , DoD is here for your review.

Just a note here: we have decided after stand-up to opt out of grouping some "demos" as "unmaintained" as this issue alone requires more careful consideration, and we don't want it to disrupt the migration of new samplers. Nevertheless, I record this issue at #696.

[jakobsj]

@chaozg DoD provides a good overview and is approved - thanks!