Rework the algorithm/environment classes and constructors

[RemyDegenne]

AI disclorure: the new list of names and plan for the organisation of algorithms and environments were done manually. The changes in the code to follow the plan (mostly renames) are done with AI.

See the new naming-convention README at LeanMachineLearning/SequentialLearning/README.md . It gives the current names for algorithms or environments that don't depend on some parts of the problem, as well as proposed names for not yet implemented classes and constructors (marked MISSING).

• Add Markov algorithms, that depend only on the observation (the state in an MDP) but not the history.
• Allow oblivious and stationary environments to have non-Unit observations. The old Unit-restricted constructors are now Environment.bandit and Environment.banditSeq.

Renames:

Kind	Old	New
Class	IsDeterministicAlg	Algorithm.IsDeterministic
Class	IsDeterministicEnv	Environment.HasDeterministicFeedback
Class	IsObliviousEnv	Environment.IsOblivious (meaning changed, see above)
Accessor	nextAction, actionZero	Algorithm.nextAction, Algorithm.actionZero
Accessor	feedbackFun, feedbackFunZero	Environment.feedbackFun, Environment.feedbackFunZero
Time zero	Algorithm.p0	Algorithm.policyZero
Time zero	Environment.obs0	Environment.obsZero
Time zero	Environment.ν0	Environment.feedbackZero
Constructor	detAlgorithm	Algorithm.deterministic
Constructor	randomSampling	Algorithm.const
Constructor	detEnvironment	Environment.detFeedback
Constructor	obliviousEnv ν (no observations)	Environment.banditSeq ν
Constructor	stationaryEnv ν (no observations)	Environment.bandit ν
Constructor	onlineEvalEnv	Environment.evalSeq (now built on banditSeq)
Constructor	evalEnv	Environment.eval

Notable breaking changes: the names Environment.stationary and Environment.oblivious still exist, but now mean the versions with observations. Code that used the old observation-free environments should switch to Environment.bandit and Environment.banditSeq.