Test: Per kernel language server for jupyter notebooks

[rchiodo]

Refs: #6333

• anyOS @alexdima
• anyOS @greazer

Complexity: 3

Create Issue

---

Intellisense in notebooks was originally provided by the python extension. We've made a change to have it provided by the jupyter extension so that we can have 'kernel' specific intellisense. This means if you select a kernel in a notebook, the packages that are parsed depend upon which kernel you pick.

Setup steps:

1. Install some version of python
2. Make sure you have pip installed
3. Install Jupyter extension from marketplace (it has insiders only versions)
4. Install insiders Python extension

Test steps:

1. Create two python environments using either venv or conda. Alternatively you can use an existing environment.
2. Install pandas and numpy into one of the environments but not the other
3. Open VS code in a new folder (if using venv, make sure the venvs you created are located at the same root)
4. Create a notebook
5. Add a cell like so at the top:

import numpy as np
import pandas as pd

6. Add another cell like so:

df = pd.DataFrame(np.arange(100).reshape(10, -1))

7. Try switching kernels.
8. Depending upon which kernel you have selected, there should be diagnostic errors on the numpy and pandas lines.
9. Try adding more cells
10. Make sure if you switch you don't get duplicate diag errors on any cells.