Prevents refractive index dispersions from being summed

[domna]

Fixes #110

Also keeps the DispersionSum flat, i.e. adds the dispersions to the internal DispersionSum list instead of creating DispersionSum(DispersionSum(Dispersion, Dispersion), Dispersion.

I think with this fix we could also release 0.15.1 to provide python 3.11 support and also to fix this bug in the published package.

[domna]

@MarJMue If you can come up with a better name than UnsummableDispersion I'm open. It sounds really weird to me but I could not think of something better 😅

[domna]

NonAdditiveDispersion? Sounds a bit weird, too...

Maybe it's weird because it is negating it. We could also switch and add additivity by an subclass with AdditiveDispersion, feels also a bit cleaner to me.

Edit: See #114

[domna]

I found a better way. I will implement it as IndexDispersion instead of UnsummableDispersion. In IndexDispersion only the refractive_index will be set and the dielectric_function will be calculated by a parent class method. This way we may also implement adding for n-based dispersions.

[MarJMue]

I think the implementation is good as is. And we can easily expand if needed. 👍