Isar proof reconstruction: type variables introduced by the ATP should be treated as type parameters for type inference purposes;
this solves the unification exception that consistently was thrown for Cezary's "FSet3" theory
failure of reconstructing the Isar proof (e.g., exception) should not prevent Sledgehammer from showing the short one-liner proof -- but in "debug" mode we do want to know what the exception is