Again that's not the fault of juries. Opinions aren't set in stone; they are prone to change. If you find that hard to swallow, then ask yourself: how often has your opinion changed between semifinal and grand final? If it never has it might indicate that your preferences were determined before the live show even began, which is the opposite of what we want in juries.
You're absolutely right that people are allowed to change their opinions! That being said, when a song jumps from one extreme to the other between the semi and the final, it does seem a bit odd. I know that some of that probably has to do with it being in a different field of songs, but I just have a hard time believing that almost all of the 16 songs added into the mix come finale would come either higher or lower than a song already graded.
I'm realizing my point is sounding a little confusing, so I'll use one of the actual jury votes as an example. Let's look at

juror #1. In semi-final 1, they rank

first out of 16.

doesn't qualify for the final, but

does, so the juror has already heard 10/26 songs in the final. Come finale, they put

25th, second to last. The song ranked last is

, which was one of the 10 songs they already knew from the semi. This means that they thought that
every single one of the 16 "new" entries was better than their previous favorite AND 8/10 of the qualifying songs from the semi were now better. This is just weird and goes against common psychological phenomena (see Mere-Exposure Effect:
http://changingminds.org/explanations/theories/mere_exposure.htm).