If the test is useless for yound dates, what grounds can we have any certainty that it works for longer periods of time?
The fact that the results for a valid test are corraborated by other dating methods.
If the tests weren't valid, they would return radically divergent dates. Instead, for objects within the valid time frame, dates from unrelated dating methods tend to converge. It's like being weighed on a spring scale, a doctor's balance, and by submersion in water. If you get the same weight each time, you can be pretty sure those methods were valid.
So no, that's not evolutions rejecting evidence - that's the realization that measurements have practical limits, and when you exceed them, you get information that doesn't match other measurements.
What he thought was the human fetal gill turned out to be our ear.
This isn't even a true story.
The human jaw
does form from an embryonic structure that, in some other organisms, becomes a gill. So Hackel wasn't a liar, but either you are, or whoever told you that story was.
Seems to me changing the definition of a word in the middle of a discussion is the same thing as throwing evidence away.
Seems to me that presenting half-remembered or outright inaccurate stories as evidence is the same thing as lying, but that's just me.
Again, you've presented a story that's just plain wrong. Evolution simply is a change in allele frequencies, which causes speciation under specific circumstances. So it's not changing the definition - evolution means both of those things, because they're the same thing.