Refactor code for calculating the number of entries

[mastricker]

I am not questioning the numbers or intend. What I am trying to say: can we refactor this piece of code in terms of the individual contributions (guess that is what @max-veit means with n, lambda, mu, ...) and then make it look like

n_entries = n + n' + lambda + mu + some_other_value

To be clearer on what is being done here.

Originally posted by @mastricker in https://github.com/cosmo-epfl/librascal/timeline

[mastricker]

As described in the references comment above. I am not happy with how the indices/columns are computed. Can these be recast in a clearer manner so that the sum in the end is as in the comment above? It also seems to me that there is some float/int implicit conversion going on, which a I am not sure.

here is the original comment in an unrelated PR.

Edit: This is an issue for a new pull request, of course and should be dealt with separately than the one referenced. And please do ask, if you have any questions.