Added tutorial on using torch optimizer with heyoka

[seba0018]

I added to the already existing tutorial "torch_and_heyoka.ipynb" a section on using the Adam optimizer from torch. I use a simplified case of the Neural ODE training tutorials to showcase the difference between simple fixed learning rate stochastic gradient descent and training with the Adam optimizer.

Extrapolating this tutorial to other torch optimizers and the use of schedulers should be straightforward.

[bluescarni]

Hi @seba0018 and thanks for the PR!

There's a somewhat cryptic warning that makes the documentation build fail:

https://app.circleci.com/pipelines/github/bluescarni/heyoka.py/1528/workflows/4657b0c1-081d-4629-8649-a724da0b52e5/jobs/4052?invite=true#step-102-643469_128

I think that the problem is that there is a wrong usage of the # symbol for introducing new sections in the markup cells of the notebook. The error message seems to be saying that you cannot follow a # (new section) with a ### (new sub-sub-section).

[seba0018]

Thanks for interpreting the error @bluescarni ! I removed all sub-sub-sections, hopefully that will resolve it. :)

[bluescarni]

Cheers! I will go ahead and merge this.