It indeed looks a lot like a celtic knot, more exactly, like a very simple quarternary celtic knot (a knot with 4 predominant corners)
Besides the obvious theme of eternity or never ending cycle celtic knots share, a 4 part knot is usually associated with meanings like:
- Four cardinal directions
- Four stations (that could make some sense with the flower in yours)
- Four of the eight Wiccan/Celtic Sabbats (Samhain, Beltaine, Imbolc, Lughnasadh the Fire related ones I believe)
- Four elements (as a magic seal/in a videogame context this is the meaning that would make more sense to me)
These knots existed way before the 12th century, as they are found in celtic art way before Christianity reached there (around, the 5th century, I believe) but they became very common on Christian art, so it probably wouldn't be weird to find them on stained glass.
I'm back, baby.
All of this seems plausibly accurate, mostly because the general shape/linear directions/usage of space and form of quarternary knots aren't all necessarily cross-shaped. Some are also X-shaped, and given the Celtic origins the X shape has also been used as a connotation to the clover. Below are instances I've amassed wherein the shape/linearities fall under this category, as well as instances of stained glass to support the portion of your inquiry relating to stained glass.
That multiple-knots image also features several knots which also fall within the above stipulations, but I don't think it's necessary for me to mark out the (pretty simple) shapes in red and blue more than a few times--since the simplicity speaks for itself, after all.
All in all, quarternary knots seem to be the most plausible, and have the necessary details and history to ring true of your two stipulations of "stained glass" and "12
th century or earlier".
EDIT: It may also be possible that it contains slight reference to the evil eye--look at the "arms" of the cross singularly and it's not hard to see how much the symbol looks like two eyes sharing one unique pupil (heh...not hard to
see). But I have no supporting evidence that even remotely backs this up, so it's purely conjecture. Neat conjecture, but conjecture nonetheless.