Industry 4.0 means a deep transformation for all our industries. Smart factories are a key feature for this transformation. Planning and scheduling are within the core functions of those smart factories. Although scheduling problems have been studied for long, the deployment of digital scheduling solutions in industries is still an ongoing process. In this work, we discuss the detailed scheduling problem in an aeronautical assembly line. It has the structure of a challenging multimode Resource-Constrained Scheduling Problem with incompatibility constraints, a resource leveling objective and also a high number of tasks. To begin with, we present two new event-based mixed-integer linear programming formulations for this problem. Then, a constraint programming formulation is also detailed. A large-neighborhood search approach based on constraint programming and tailored to the resource leveling objective is presented. The four approaches are tested and compared using \emph{ad hoc} data sets and industrial instances, yielding significant improvement compared to the heuristic currently used by the company. Moreover, the large-neighborhood search approach significantly improves the method proposed in the literature on a related multimode resource investment problem when short CPU times are required.