I simply considered the looking glass as a mirror...
No, It is a lens ! In a lens, the "right" side of an object remains "right" on the plate where it is focused, behind the lens.
As I had noticed, there should be catadioptric "looking glasses" / telescopes, which do mirror transformation. I do not know anything more on that subject - I had never used one of those light catadioptric tele-lenses on a camera either ( although they have two mirrors - so I guess they reverse the reflected image two times, and so, in the end, they do not mirror-reverse it at all. )
I think we have to consider the tying process (how we tie the knot). If we consider the tying process, the two knots (the "real" one and its reflexion) are not the same knot: one is righthanded and the other is lefthanded.
Noope !
Take a picture of a right-handed bowline, and a picture of its mirror-symmetric, and post both of them in the Forum. Say that the one is a right-handed, bowline, and the other left-handed bowline. Listen to what you will listen...

THAT is my point : the handedness ( of their nipping loops ) changes, but the practical knot ( the bowline ), as such, does not.
Moreover, the tying process is enough for some knots, as you say, but not for all ! Starting from the Double bowline on-the-bight, and extending both strands of the collar after its tip, without altering the side of the Standing/Tail Ends pair from which the two lines have already passed by, ( so, by transporting the collar, so it collars another pair of segments, further inside the nub ), you can tie more complex Med bowlines, for which we can not say if they are left-or right-handed - simply because we can tie them both ways, by first forming the one or the other single bowline and then retracing it line. It is the fact that the "parent" of all Med bowlines, the Double bowline on-the-bight, can be considered as either a left-handed or a right-handed bowline, that puzzles me. How the "parent" does not have one, only, defined handedness, and the "children" acquire one, and one only ? From which, exactly, point of the double line and afterwards, do the side of the segments around which the two lines of the double line pass by counts ?