Should there be a vectorisation example notebook to explain how broadcasting works?

For the simple.ipynb notebook, we will have to wait for #213 to be merged.

The gradientbased.ipynb notebook should be rewritten to take advantage of the vectorisation. I think this kind of goes hand in hand with rewriting the same example in the cheetah-demos repository.

The Accelerator, Error, Track Methods and Utils doc pages are empty 🤔

In Particles, only the docs for Beam appear. ParticleBeam and ParameterBeam are missing.

I think in gradientbased.ipynb, where we show how to use the gradients for accelerator tuning, we can't really take advantage of the vectorisation, because we always want to use the gradient at the current magnet settings(?)

Because their branches are already merged in here, this PR MUST only be merged into master AFTER #213 and #265.

The Accelerator, Error, Track Methods and Utils doc pages are empty 🤔

In Particles, only the docs for Beam appear. ParticleBeam and ParameterBeam are missing.

So it turns out that rendering the docs does not work for files that have relative imports in them. We will have to keep that in mind for the future. For now, I returned all imports back to absolute imports (except for those in __init__.pys).

Because their branches are already merged in here, this PR MUST only be merged into master AFTER #213 and #265.

‼️

The Accelerator, Error, Track Methods and Utils doc pages are empty 🤔
In Particles, only the docs for Beam appear. ParticleBeam and ParameterBeam are missing.

So it turns out that rendering the docs does not work for files that have relative imports in them. We will have to keep that in mind for the future. For now, I returned all imports back to absolute imports (except for those in __init__.pys).

On a side note, what was the reason to switch to relative import? I think we started with having the full imports anyway.

The Accelerator, Error, Track Methods and Utils doc pages are empty 🤔
In Particles, only the docs for Beam appear. ParticleBeam and ParameterBeam are missing.

So it turns out that rendering the docs does not work for files that have relative imports in them. We will have to keep that in mind for the future. For now, I returned all imports back to absolute imports (except for those in __init__.pys).

On a side note, what was the reason to switch to relative import? I think we started with having the full imports anyway.

I thought we started having the relative ones 😄. In general, I would argue that relative imports are better Python style.

The issue seems to be that Sphinx doesn't like them. The annoying bit there is that this error happens mostly silently. I only noticed it because parts of the docs were missing (specifically those from files that had at least one relative import in them). The only real clue that this was happening was in the build logs on readthedocs, where if you scroll through and look closely, you can find warnings like this one

WARNING: Failed to import accelerator.aperture.
Possible hints:
* KeyError: 'accelerator'
* ImportError: attempted relative import beyond top-level package

And then of course the fix is to make all imports absolute, e.g.

from  cheetah.themoduleyouwant import thethingyouwant