This needs to be tested by the ISVV team, not by players, because it requires very large sample with the same weapons, then with different weapons, and using the same way of collecting data to be able to draw the most accurate % and/or PPM.
Right now, depending on your sample sizes made with the same weapons, you can get results which vary from 4,5% to 10% depending on rng. (So between around 1 to 3 PPM basically).
How reliable would you (and the devs) consider the 1 PPM from the EJ thread ? We don't know what kind of testing and its conditions lead to this player-made conclusion.
Maybe he just took 1 PPM as the "worst case scenario" for modeling purposes in his sheet, we don't and can't know.
You can read in it that different rogues think it is worth 8, 10 or even 15 dps in raids. Its value probably varied a lot depending on the players' raiding environments, the sample sizes and weapons used as well.
You can read elsewhere in later-than-vanilla-patches people stating it's around 2,8%, another guy states 3%. Again, we don't know with what kind of weapons and under which circumstances.
1 PPM seems to usually be the value for character "enhancements" type of procs, like Crusader.
For procs which are directly dealing damage and/or apply a debuff, it is usually higher than 1 PPM, like Fiery Weapon & Lifestealing.
I would think BF falls into the 2nd category. So >1 PPM is what it should be.
Moreover, things which are PPM based are most of the time related to weapon enchants. The rest is either PPM or can also be flat proc rate (like HoJ & sword spec). Why wouldn't BF be a flat proc rate instead of PPM ? Is there any 100% reliable source saying it's PPM based even ?
The fact people would think/feel it's broken here could also be explained by external factors, such as the "chance on hit" mechanism working for misses/dodges/parries/etc., which is a reported & confirmed bug. So as long as this bug isn't fixed, you can't draw any conclusions regarding BF proc as a player doing couple testing samples in raids.
This bug alone is making every testing samples about "chance on hit" effects non accurate at all.
Then there is the difference between having 1 PPM in mind, and how you will feel it in game when you are tossing a
"downright i n s a n e" amount of special attacks under a set period of time which will make you get a
"downright i n s a n e" number of procs depending on luck and lead to such different results.