Restraining the definition to knowable limits makes it less dogmatic and more universally accepted.
The problem with this position is that it must be applied to all definitions and it is unnecessary. If the allele's of a population change, it is evolution. If they haven't changed - it isn't. It would be like saying murder is unlawful killing that evidence demonstrates occurred. It isn't - murder is any unlawful killing (I appreciate there are other conditions depending on the legal system, but none of them change the point I'm making). Whether we can say a murder has happened in any given case depends on the evidence.
Likewise, whether evolution happened in any given case depends on the evidence.
One must be open minded, and this how being open minded works (since I am sure you consider yourself to be so, I doubt you will argue this):
Assume natural history as described by scientific consensus happened - what would we call the process that generated this history? We would call it evolution.
The question of whether natural history happened the way they say it did or not is irrelevant to the name we would give the process that led to it if it did.
What's wrong with seperation of definition and religious inference?
The definition of evolution make no inferences whatsoever. The definition of evolution does not infer natural history in any way shape or form. The definition of evolution does not even infer that it happens at all and nor should it. The definition merely states that if a change in allele frequency happens, then biological evolution has occurred.