The Mystic has Int 5, after 3 xp she had Int 5 + 3 fortunes. Int governs 9 out of 26 (tradecraft is left out) skills in the game, with some very important ones like First Aid or Observation. You can do a few numbers and see which is the check success ratio for someone with this stats, 65% for a Hard check and 80% for an Average check, and if you do a bit more numbers you will see that this means that with a merely 3xp this player has boosted his success ration in 9 (ok not 9, because some are advanced) skills by 20%!!. I have always found fortune advancements without regulation totally game breaking.
Try to ask PCs to branch out in WFRPG 3 when you have a party of 8 players and they have 20 xp.
I understand I amb the GM and I have to be creative about resolving conflicts and that no system is perfect, but in wfrpg 3 I start to feel I have to be creative to the point of inventing my own game system. I have paid a huge amount of money for a game and I want a finished and play tested product. And for sure I am not going to punish my players for being "munchkins", because this is a rather natural thing to do in a game. If you are a fighter, you want to be strong, good with the sword and kick the ass of as many orcs as possible. I would rather prefer that ffg has produced a more solid system (like others in the market) where the possibility of behaving as a "munchkin" is rather complicated or impossible at all because the system does not promote /allow it.
The other day a player of my group did a reflexion that left me a bit shocked. The comment arised after playing a game where his 10xp character, was able to convince with charm and guile half of the population of Ubersreik to lick his boots. He said "In this rpg PCs start the game as real bad asses, and they progress really fast, with 10xp they are near heroes. On the other hand combats are very deadly and it is easy that you die. It looks like it is made to play for a few sessions and hoops, create a new PC. It remembers me of the concept of boardgames like Advanced Hero Quest".
Well, sure, the Mystic will be really really good at those intelligence checks. But how is his fellowship? Or physical stats for that matter. By punish I didn't really mean that you should do this in a meta-game sense. I meant more that you should put him in situations where the intelligence is not the stat to check. Of course, sometimes you want the player to have a chance at shining, then it's fine. That aside, try using more opposed checks where the opposition is someone with at least 5 in the opposed ability. Perhaps his observation check is not always vs Stealt(Ag) but could be vs Stealth(Fel) for someone trying to hide in a crowd (if the NPC has a higher Fel score). Or perhaps his Observation check is not Int, it could be argued that you have to use Willpower for long stake-outs, or Fellowship for noticing aberrant behaviour. Use the ACE budget liberally. Almost never allow the players to use stance dice in story mode.
Using http://laakmann.free.fr/wfrp/ with a pool of 5 blue, 2 yellow, 3 white, 2 black (from opposed training or cunning dice) and 2 purple the prob. of success is 83%. This looks perfectly alright to me, the mystic should succeed often in the area he specialized in. Without the white dice the success rate is 70%. The precise effect of 3 fortune dice will vary with how the dice pool looks, but 13% increase for 3 XP does not seem to bad for to me.
I can totally understand the problems with branching out and having 8 players. But the game isn't really designed for that many players, I'm not surprised you have to use houserules to make it managable. I can't really think of a RPG which would not have these issues though, then again I've never played in such a large group. The large group you have probably also affects how easy the published adventures are since your players can specialize without any risk, other players will always be able to cover for any lack of branching. In my 3 player group, even the wizard has been thinking about taking a non-wizard career just to cover some of the skills the party lacks.
One easy way of making the skill checks harder is to demand 3 successes for a complete success and that you only get a partial success with 1. The prob in the hypothetical situation above then shrinks to ~50%. I used this kind of method when running an investigative adventure and it worked really well.