CI/TSML: Fix Time Series builds by downgrading to Python 3.11.8

[amotl]

Problem

GH-403 fixed a few flaws when running Jupyter Notebooks on Google Colab, as reported by @marijaselakovic. She reported the Time Series slot also has flaws, this patch aims to mitigate them.

Solution

We are fixing this on CI/GHA by downgrading to Python 3.11.8. On Google Colab, it will work anyway because that environment is still running Python 3.10.

References

More information about why Python 3.11.9 breaks Dask:

• [BUG]: TypeError: descriptor '__call__' for 'type' objects doesn't apply to a 'property' object pycaret/pycaret#3960 (comment)

[amotl]

Problem

ValueError: Found array with 0 sample(s) (shape=(0, 1)) while a minimum of 1 is required by SimpleImputer.

-- https://github.com/crate/cratedb-examples/actions/runs/8631971131/job/23661528515?pr=404#step:6:910

[amotl]

The ValueError coming from SimpleImputer might still be present, and could be an unrelated fluke?

-- https://github.com/crate/cratedb-examples/actions/runs/8632142024/job/23662092328#step:6:1499

Re-running...

-- https://github.com/crate/cratedb-examples/actions/runs/8632261164/job/23662506620