Attack Overview with Properties

The table at [foolbox.attacks](https://foolbox.readthedocs.io/en/stable/modules/attacks.html#module-foolbox.attacks) is currently the closest thing to an overview of all implemented attacks. It could be an ideal place to list major defining properties, such as whether an attack minimizes perturbations and if it supports targeted attacks (or rather, which criteria it supports).

1. Is it correct that no list with more detail exists?
2. What other properties should be included?
3. Could/should the overview be generated automatically, possibly by inspecting which classes (such as [`MinimizationAttack`](https://github.com/bethgelab/foolbox/blob/92af8673ee957e5af1f9ad5f3abb474c63f93397/foolbox/attacks/base.py#L334)) a given attack inherits from?