Kev, I'm not sure there is an extra burn from mirror 1 or that is just dirt on the test card.
Agron, I believe that mirrors 2 and mirror 3 are identical. (If I remember correctly mirrors 1 and 4 have heat sinks on the back of the mirror plate.) Maybe you could interchange mirrors 2 and 3 and retest? (Or swap in different mirrors if you have any.)
If you are in a pinch and don't have spare mirrors you might be able to cut some "mirrors" from an old disk drive platter just to test the optics. They don't have to be circles - you could shear rectangles from the aluminum disk platter. I am not recommending this permanently, but just as a low-cost test if you don't have a budget to buy parts to diagnose the problem. Tradeoff between time and money. Using old disk drive platters to make mirrors has been discussed in separate threads.
Also, when you test, perhaps you could reduce the power and/or burn time. Set the power to maybe 10 or 20% and turn the laser on for a second or less (whatever it takes to make a tiny clean burn mark.)
It seems like there is a thermal issue going on during vector cutting - the heating effect seems to be causing shifting of optical alignment. So maybe mirror 2 or mirror 3 is heating up too much and warping the mounting system. Have you checked if either of these mirrors and mounts are getting hot? You might have to pull them out of the mount and touch the back to tell, unless you have some other way. (But keep in mind that removing and replacing mirrors can affect alignment.) You could also try to feel the back of mirror mount #1 to see if it is overheating.
I suspect that raster engraving isn't heating the mounts up as much so you seem to be able to get away with it.