It does seem odd that the existence of Mpembra has been debated for decades and it's not yet settled. (I first heard about it decades ago and I know little about it other than it was reported as one of the phenomena of water that was still not well understood.)
Just a thought, if the effect is simply an statistical/emergent phenomenon like temperature then the many parameters involved, initial conditions, bulk amount, physical setup etc. might be the reason why repeatability is inconsistent. Perhaps there have not been enough rigorous studies.
Also, one wonders about the accuracy of reports such as this when errors of this magnitude get through the editing process (the date of ref [1] is 1969):
[1]. The phenomenon has sparked intense debates for more than two millennia and continues to do so [millennia?]