porting dsci 100 nbgrader notebooks to python

[phaustin]

As part of the python port for the dsci 100 R notebooks, we'll need to move all nbgrader tests to the python kernel.

• [ ] install and do the tutorial at https://nbgrader.readthedocs.io/en/stable/
• [ ] review current asyncio, e.g. https://realpython.com/async-io-python/
• [ ] review new nbclient notebook execution api: https://github.com/jupyter/nbclient/blob/master/nbclient/tests/test_client.py
• [ ] get up to speed with the changes made at https://github.com/UBC-DSCI/nbgrader/tree/generate_tests_stringeq
  • see https://github.com/eoas-ubc/eoas_tlef/blob/master/docs/nbgrader.md
    • [ ] Rewrite _execute_code_snippet to use code from the new nbclient.NotebookClient
    • [ ] Also finish off the related limit_testcell_height
• [ ] help prepare pull request to upstream

[trevorcampbell]

Some minor additional comments:

• I'm not sure about test_client.py, but NotebookClient shows a low-level interaction with the kernel. You may need functionality from there to run individual lines of code (but if there's a higher level API to use with examples in test_client.py, definitely use that instead)
• take a look at the ExecutePreprocessor which is what the InstantiateTests preprocessor is based on
• we want to rewrite _execute_code_snippet using asyncio to conform to the new nbclient and nbconvert code. You'll see a big block of functions after the comments in instantiatetests.py -- all of these are just there to execute individual lines of code
• the limit_testcell_height branch makes modifications to the nbgrader notebook format to add a property max_height to cells. I developed this because the autotest code was getting really long and ugly, so limiting testcell height helps keep things clean. This branch should be PR'd at the same time as autotest and the two should be cross referenced. I believe this branch is already in really good shape -- I don't remember wanting to change/add anything -- but you should test it thoroughly before PRing

and some additional todo items Before starting on the coding things in @phaustin 's message above:

• [x] pull changes from the master branch of https://github.com/jupyter/nbgrader into the master branch of the forked nbgrader repo
• [x] rebase the generate_tests_stringeq branch on that new master branch

I just did these now to avoid confusion.

one modification to the code after the above coding things in @phaustin 's message are done:

• [ ] if there isn't one already, add a warning to the type dispatch when it can't find a specific test and falls back to the default generic test (probably at the [INFO] level so that instructors will see it)