The cut or contaminant can be very very potent. This way, a low proportion e.g. 1:100000 of a high potency contaminant would make it very hard to detect while still allowing it to exert its effect on the nervous system.
This is not far fetched, there are substances which are pharmacologically active at the nanogram range - for example some of the new synthetic opioids among many other potent drugs.
I'd bet a pure racemic 3,4-MDMA HCl mixed with one of these very potent sedating contaminants in the proportion 1:100000 would be pretty Meh and almost impossible to detect and separate.
The whole point of re-X is to purify the main substance from impurities, so a contaminant that would be active in the nanograms range would never find its way to the crystallized product but it would stay in the mother liquor. In this case the re-X was done 3 times.
If instead the contaminant/cut is present in quantities enough to recrystallize along with the main product it would: (A) be detected by GC and (B) increase the threshold dosage. Both possibilities are false in this case.
So this leave us with only one possible outcome: the MEH product is an active isomer.
My bet is that MEH is 2,3-MDMA or one of this table: