Missions including close-proximity operations and docking (CPOD) as well as in-orbit servicing are redefining the capabilities of small spacecraft. As a result, attitude and orbit control systems (AOCS) requirements have evolved to include not only attitude maintenance, but also 6-DoF translation and rotation. Cold-gas thruster arrays are key to these capabilities, but optimising their design for sustained as well as impulsive performance while maintaining redundancy cannot be addressed by existing impulsive-centric methods. In this work, we present a coupled analytical and simulation design framework for defining thruster configurations while imposing deterministic redundancy. Established convex geometry methods for fault-tolerant design are reformulated to always ensure sufficient control authority, dictated by close-proximity operations and docking capability. A modular simulation stack allows for the inclusion of a Reinforcement Learning policy as a CPOD guidance module and of effects such as valve transients and small-scale nozzle efficiency. An optimised thruster layout is obtained using a genetic algorithm and verified for a rendezvous and servicing mission, representing an in-orbit servicing demonstration of WEP The observed 15.7% reduction in thruster on-time is contextualised against typical performance drivers in cold-gas AOCS design. This work establishes a baseline for AOCS hardware development within the Ice2Thrust EIC Pathfinder project. Ice2Thrust proposes WEP as a non-toxic, high-performance, and refillable alternative to conventional propulsion systems. By integrating redundancy and realistic system dynamics early in the design process, the present approach contributes to a system that is robust and prepared for mission extension applications.