Given that ~100 million tons of oxygen are produced annually, extracting all the xenon from that air would yield about 170 tons/year. So there's a bit of room for growth.

The BepiColombo number is similar, I think, to the amount of xenon made annually in nuclear reactors (where it occurs in spent fuel as the result of fission.)

I think it might have taken a larger percentage of high-grade ultrapure xenon, a narrower market than the global bulk supply. A 1% impurity is fine if you are using xenon for welding, not so much if you are firing zenon plasma at a grid carrying a few hundred volts. A little bit of o2 in there and your grid would be rust very quickly.