Essentially, you assume here that the distribution of "universes someone bothers to simulate" is different from the Solomonoff priors in a way that makes universes where religion is true more likely.

May be. If the simulators are humans, some of them would enjoy playing gods; and even if it is a small minority, it would still be a larger fraction than universes where gods "naturally" exist as complicated laws of physics. If the simulators are alien intelligences... well, I would be less certain about those, but still seems like the fraction of situations where the simulator decides to personally interact with the simulation should be larger than universes where an intelligence is "hardcoded" into laws of physics.

But there would still be many simulated universes where the simulator is not interested in interaction with sentient beings, and all religions arise naturally for reasons unrelated to their correctness. Or the simulator would interact with the world, but in a manner totally different from what religions talk about; imagine for example that our world is just a computer game played by a bored teenager who once in a few millions of years clicks a mouse button to drop a huge meteor and change ecosystem; the individual humans are too small and short-lived for him to even notice. Maybe the simulator completely ignores humans, and is only using this universe as an incubator for an AGI that he will later ask to compute some mathematical problem.

So, the probability of some religion being true could be greater, but still relatively small. Not sure if large enough to excuse Pascal's wager.