Fixes bugs in nn interpolation with gradients

[yvind]

This PR fixes #130 by implementing the suggested changes for alpha and beta.
As well as the suggested change for diff.

The doc string is updated with a note on keeping flatness >= 0.0 and that flatness = 0.5 reproduces the original Sibson C1 interpolant. As well as a refrence to Julia Flötotto's thesis and the equality of flatness=0.0 and Flötotto's I1 interpolant (and it's connection to Sibson C0 interpolant)

[Stoeoef]

Can you also please run cargo run --example interpolation and commit the changed images? This will update the images in the help appropriately.
I have also observed changes to the the barycentric interpolation image for some reason. Feel free to include it or leave as is.

[yvind]

Will do! I don't know about the barycentric interpolation as I haven't touched anything outside the interpolate_gradient function

[yvind]

I didn't observe any chenges to the barycentric images. Only the interpolation_nn_c1 images changed for me when running the interpolation example.

[yvind]

Hi! Are you willing to merge this or do I need to do something I have missed?

[Stoeoef]

Oh dear, my apologies. I forgot about this.

All fine from my side - please ignore the clippy failure, this can sometimes happen if CI uses a newer version of clippy. Merging in!