Add CT experiment demos with info about when to use them

[AnderBiguri]

CT experiments are great, but we need to communicate well to mathematicians who just want to try their models in CT which ones are easier, harder, more interesting, more challening etc.

[mbkiss]

I would propose to add the final description of the "experiments" or CT reconstruction tasks as in the final version of the benchmarking paper. This should give an intuition on what happens in these CT experiments. The difficulty of the experiments is algorithm dependent, so I would not make a generalized statement for all tasks but rather say in general that in with increasing sparsity and limitation of angle it gets more difficult.