Add Fixes for Post-Interrupt Hook Feature

[piiteraq]

This is a rework of #1186 to make the post-interrupt hook feature work as intended:

• Added abstract functions to ‘JupyterApi.scala’ to make interface visible to Ammonite kernel object
• Added API usage examples to ‘api-jupyter.md’
• Added convenience functions for facilitating JupyterApi instance access,
  since it is hard to pass around through function/method calls. In case at
  hand, we need to access it from methods ‘interruptible’ and ‘interrupt’ in
  class ‘Execute’
• Added a label field to each registered hook, since Scala lambda functions
  are not easy to print in human readable form
• Did local testing of API both when sending interrupt messages via SIGINT or via
  zeromq message, i.e. without/with '--interrupt-via-message’)

Note: Contrary to execution hooks, interrupt hooks do not at present funnel output
from one hook into the input of the next in the chain.

Please find screenshots from local testing attached below.

[alexarchambault]

@piiteraq Going to merge and cut a release before the end of the day Europe time. Just tell me if you have an opinion about the extra commits I pushed.

[piiteraq]

Hi @alexarchambault - Thank you so much for the review and the fixes - I agree on all of them. I had difficulties figuring out how to pass the JupyterApi instance down through the call chain through normal parameter passing, so added the instance setter/getter and thought you might have a better idea, which indeed turned out to be the case. Also, you fixed the problem I had with accessing the logger in the proper location. Thus, please proceed with the release. Thx --