user-defined environments in jupyterlab

[mishaschwartz]

Right now users can use one of the pre-created kernels (environments) provided to jupyterlab.
It would be nice to allow users to define their own package requirements using either conda, pip, etc.
This will allow for users to define specific package versions that they need for their workflows and will ensure that notebooks/workflows don't break if there is a breaking change in a package dependency.

We should make sure that this is available for both python and R kernels