Better handling of empty Hoeffding trees

[rcurtin]

In #2962, @xlindo pointed out some issues when trying to train a Hoeffding tree after it was created with an empty constructor. I went through the code and realized that there are some situations where this does not work correctly; therefore, I went through and refactored things such that the code in #2962 now works. In essence, the problem was that a HoeffdingTree<> stores a data::DatasetInfo and a few other parameters that track how many dimensions are in the data and what type those are, but the code does not provide clear facilities for how to modify those quantities.

Specifically, I made the following changes:

• I added a bunch of documentation to the overloads of Train() to be clear about when the tree will and won't be reset.
• I added a specific parameter resetTree to the overload of Train() that doesn't take a datasetInfo. This makes it possible to reset the internal state of a tree even without passing a DatasetInfo.
• I refactored training and resetting into TrainInternal() and ResetTree(), which are meant to be called internally.
• I added some new test cases to ensure that the original code works now.

[zoq]

Looks great to me.

[mlpack-bot]

Second approval provided automatically after 24 hours. 👍