Learner at entrance and exit tasks

This is from the work currently going on with A.Shaow and M.Dotz. 

Learners whose loops are hard-coded require pre/post entry and exist hooks for pre/post processing.

Example:


```python

@al.on_learner_start
async def pre_process():
       return

# ........ simulation/training/etc

@al.on_learner_finish
async def post_process():
       return

```


