As I see it, a modular fighter would be modular at the factory, and not particularly modular in terms of field reconfiguration. The main feature that changes between versions is the weapon package: missiles, lasers, fusion, particle, boarding, etc. A lot of those have different energy requirements; a missile fighter would have a lot of resources wasted on power plant if it used the same power plant as a fusion fighter. Rather than waste the money on excess power, a missile fighter would do better to add magazine space or increased maneuver agility (if the latter were possible within the G limit for the fighter's technology level). So rather than a swappable module design, it would be a class design that benefits from common design and manufacture discounts.
The other reason to rule it that way is less a "rules make us do it this way" than analogy with real life. You don't just stick another engine in an F-16 and make it an F-15. Even the absurdly expensive F-35, which is meant to be built in assorted versions, isn't meant to allow a VTOL version to go into the shop one day and come out as a carrier landing or long runway version a while later. Each version is optimized for its role, and doesn't change role for the lifetime of the fighter.
So, I would propose to make it a class of factory modular fighters, each version the best it can be for its role. If a fleet has the wrong mix of versions, combat attrition will allow the factory to realign the version mix by building different variations within the class.