Add functions to make proof reporting easier

Add functions to make proof reporting easier

solver_get_learnt: return all rearnt rules used in the proof
of a problem or another learnt rule.

solver_get_proof: return the decisions that led to the
problem or learnt rule.

Note that this is experimental and subject to change. We will
need to add some sorting code to solver_get_proof() to group
similar steps, but this means that we need to get the
ruleinfo for the involved rules. But in that case we can
as well also return the ruleinfo data with the call.