Botorch with cardinality constraint via sampling

[Waschenbacher]

This PR attempts to support cardinality constraints in botorch recommender. The core idea is

• Determine several configurations of inactive parameters (parameters that have to be zeros) based on the cardinality constraints.
• Optimize the acquisition function for different configurations and pick the best one.

There are two mechanisms for determining the configuration of inactive parameters.

• When the combinatorial list of different inactive parameter configurations is not too large, we iterate the full list;
• otherwise a fixed amount of inactive parameter configurations are randomly chosen.

In addition, a validation is added to ensure that zero is included in the bounds of all cardinality constraint parameters.