Change default Python version to 3.8

[astrojuanlu]

Describe the feature you'd love to have in Orchest
Many packages from the scientific ecosystem follow NEP 29 - Recommend Python and NumPy version support as a community policy standard. Examples include pandas and other libraries.

NEP 29 encourages library developers to drop Python 3.7 as of December 2021 (couple of months ago), and so both NumPy (numpy/numpy#19665) and pandas (pandas-dev/pandas#41678) followed suit. It's just a matter of time that the latest versions of these libraries don't work on our default Python version anymore.

Python 3.8 will be supported by libraries until April 2023. We could as well adopt the same policy for our images.

Is your feature request related to a problem? Please describe.
I sometimes want to use the most up to date versions of PyData libraries, and for that I need to upgrade the default ones. For example, some colormaps like turbo are only available in matplotlib 3.3.0 onwards.

Additional context
This is somewhat related to gh-763, in that we are using an old image because of the elyra Docker baseline, itself based on an old version of jupyter/scipy-notebook.

[yannickperrenet]

Great ticket to pick up when working on #760.

For the near future I think we should be able to change the Python version that is running on the default conda environment. Basically removing the conda environment and adding a another one with a more up-to-date version (example from our docs).