Estevao, thank you for your "mini-study"! And I agree with your conclusion. Unfortunately, Wolfram's data may be inconsistent even within itself different data functions may provide different values for the same entity. E.g., when using functions like Quantity, Entity (with "Particle"/"PhysicalConstant") or ParticleData you get inconsistent values for the proton's mass or elementary charge, which seems to affect the results of UnitConvert function. I came upon it when trying to calculate the mass/energy change in a hydrogen atom formation reaction. So, currently Wolfram's data cannot be relied upon.
P.S. What is the $$Data variable supposed to denote in your code? Can you provide a downloadable version of your study (as an .nb fiile)?