In that specific case presented actually ‘yes’.
In that case it would cause ‘fun’ for a small subset of players while invalidating the ‘fun’ of the others.
That’s a immediate Day 1 fix situation even.
Very severe and shouldn’t ever happen.
On the upper-tier it was detested actually, it went against class balancing in a major way making paladins absolutely OP in comparison to what they were supposed to be.
Yes, the interaction itself was very fun, but the outcry of ‘don’t change it’ came primarily from paladin players… which… obviously say ‘don’t take away the stuff making is better then everyone else’ 
I mean… kinda obvious that’ll happen.
For everyone else? It was a extremely bad decision.
Also not very good to have. Albeit can at least be pushed into ‘yeah, it’s the current meta’ as it does primarily affect PvE environments. In PvP that wasn’t all too important, but it still was a major downside as to why Blizzard lost a boatload of long-term players during BFA and after. WoW always had balance issues… but Blizzard generally did stuff to alleviate it after a while, unlike in BFA where they didn’t give a single shit and disgruntled many many of their influencers along with it, big ones too.
You had to give up something for it though, which made it a high-stakes move to build for it. Something like that is viable. More so especially after the cap.
In some of the other examples that simply wasn’t the case, pure upsides without downsides.
The issue is that most devs don’t, Blizzard is very very prone to mishandling those things, so the examples from WoW usually are a bit of a detriment there.
A better example for ‘how to handle unintended mechanics well’ would actually be Trackmania. Which yes… is kinda weird to say any game with Ubisoft as a publisher actually does good stuff… but for that dev-studio under them it upholds even.
In Trackmania United Forever players found a good amount of bugs over the course of time. The game has a deterministic physics engine, meaning if players put in specific inputs then the outcome will in 100% of the time be exactly the same, making precision and control extremely important.
But… that also counted for the majority of bugs! These include things like the so called ‘Air-brake’ which is unintended but kept. Basically your car tilts as you jump, but if you break in the air it stops that tilt entirely. That makes for allowing timed breaks to sustain speed or even cause alternative routes at times. Since it’s entirely skill-based to do so it was left in.
Same with ‘speed slides’ of all kinds. Depending on speed when your skitmarks overlap to a certain degree instead of loosing speed you’ll gain speed. Or on ice when your car has a specific angle while countersteering you speed up rather then slowing down.
Bugslides are a extreme example of that. When your car is 90 degrees to the track after landing, allowing extreme sudden boosts of speed.
All of them are skill-based and repeatable mechanics which every player can access in a reliable way, increasing the skill-ceiling without undermining competitions for example.
Other bugs - like detection bugs for collission - do the opposite though since they’re based on tick-rate of the game, so a hitbox for one player will be determined 0,01 seconds later then for another, allowing on specific tracks for a tire to clip through the hitbox to a certain degree more and hitting secondary overlapping surfaces behind, speeding up or slowing down the car.
Those are not reliable in replicating, so they’re deemed ‘actual bugs’ rather then ‘techniques’ there.
The whole topic is quite detailed to dismantle what is good or not good, so it’s always a struggle for the devs to deal with those things when it’s not as straight-forward as in Trackmania.
Hence why it’s so rarely seen to leave bugs persistent for the future… unless you have really sloppy developers like Blizzard.
First of all… gamers are not inherently competitive, it’s a miniscule sub-set with a loud voice and visibility on the field.
Secondly… as a competitive player you want persistence and not variance. To showcase the difference between personal skill rather then RNG.
Which is why ARPGs are not in e-sports, they sheer difference in outcomes based on the high methods of RNG presented is just too large to warrant and reliable competition which would have close calls. One item drop, one bad damage roll, one unlucky combination of mobs can all change the outcome. The player can only to a degree influence their progression speed… but at the top end? When you got reliable competent players and put them all into a mid-term competition then the outcome will turn out quite different each time. The first place can easily be last… and the last easily first in such a limited field.
Unintended fun mechanics are only existent for a short time with a reason though.
One part is to not influence the invested effort into builds that need extreme effort to make them happen… and unless the unintended outcomes aren’t too extreme they’re kept up to the next big balance patch in that case.
But if they have too much of a fallout it’ll nonetheless be done immediately.
This includes things like ‘Poet’s Pen Detonate Dead’ in PoE, a very iconic and extremely fun builds… or simply overpowered builds like Dive Bomb Falconer in 1.0.
Great to have ‘for now’… but bad to have ‘forever’ so to speak.
Most of the time true.
But not always!
I get the notion though.