Wavefunctions are the usual tool in Quantum Mechanics, but many physicists (including Rovelli or Kastner) are beginning to realize that this computational tool requires a deep conceptual rework on how we interpret reality. Either we need to consider that wavefunctions collapse retroactively to handle conditions at superluminal distances, or that every possibility they describe really exists, just in a different world.
But wavefunctions, by including the observer's t and x variables don't correctly describe the true nature of the massless quanta shaping phenomena at the speed of light. At the speed of light, there's no superposition possible, because the entire spacetime interval under consideration is just a factual realization, an instantaneous event. Thus, for the massless quanta forming phenomena, nothing evolves or changes in space and time. Wavefunctions are trying to describe the instantaneous events experienced by massless quanta from the point of view of the material observer, and that's why they fall short.
All energy or information exchanges are collapsed wavefunctions at the time of detection, linking those detections to past emission events through all possible light-like paths between them, while simultaneously computing all conditions on that volume of spacetime in their point-like, timeless existence.
From this alternative point of view, the situation you describe would be as follows:
If there's only the electron gun, the double slit, and the screen (and you don't place more constraints for how light-like interactions should unfold) there are some light-like conditions, relations, and paths between the gun and the screen that can actually affect each other through both slits, increasing or decreasing the probability of some electron's interactions appearing more likely in some places than others (constructive or destructive interference).
The moment you also participate in observing the electron as it "propagates" through spacetime, you are modifying the light-like conditions, relations and paths that allow it to emerge to reality. You are changing the conditions, so there are a number of interactions that by ending on your detectors do not contribute to shaping the electron pattern as before, so there are much fewer (if any) conditions that allow different light-like paths through the slits to interfere with each other. Your detectors modify the space of possibilities for the forming electrons, changing the probabilities at the different places on the screen so that the interference pattern cannot really form.
It's like diverting a stream of water to make sure it flows, and expecting that what ends up in the bucket to be the same amount as before.
