🤕🔨 Change definition of the `alpha`-parameter to be more Fourier-friendly

[MothNik]

The previous definition of the scaling with the coefficient $\alpha$ lead to the Hermite polynomials

and Hermite functions

So, when $\alpha$ is less than 1, the function is shrinking towards the y-axis which made perfect sense (smaller alpha, smaller function).

This was okay until I realised that this was their definition in the frequency and not the time domain. Now, this is where the weird effects happen. When $\alpha$ is small in the frequency domain, the Hermite functions in the time domain are getting stretched. With this unintuitive behaviour, it's getting super confusing to select $\alpha$ properly given only the time domain. Therefore - now that there is if at all a handful of people using this repository - I need to change the definition to $\alpha{new}=\frac{1}{\alpha{old}}$ which leads to

and

[MothNik]

✔️ Checklist of things to do

• [x] update implementation within functions
• [x] remove redundant Numba legacy Hermite function and just jit-compile the NumPy legacy Hermite function
• [x] update function docstrings
• [x] update test reference files
• [x] update tests
• [x] update equation images
• [x] update example images
• [x] update README
• [x] final checkup

[MothNik]

Ready to be closed 👍