Nestable stack not used for untracked experiments

Closes #56 (closed)

This improves the logic used around identifying nested experiments when they aren't being "run". #run is technically called even when the experiment isn't tracking events or results, but this is mostly to handle control path behaviors. But we don't really consider this as an experiment run, so we shouldn't consider it as being added to the stack of experiments, and therefore, it shouldn't raise an exception if we don't add it to the stack.

The base logic comes down to if we should track events or not, and if we don't think we should track events, we aren't really running the experiment.