I would argue, that the bonuses to their max attributes are not the same as buying the "Exceptional Attribute" quality. With a metatype you are either buying the set or you are buying nothing, while the quality is a choice of the exact attribute you want. The latter to me is worth more, than the set is, while the set could by chance be the exact thing I wanted. Also the reduced maxima potentially weigh more in my book, as it is easier to be impacted by them, than it would be to make use of an increased one. Also in the metatypes they are again part of the set.
So to me it appears that the list of qualities is actually not a good toolkit to build metatypes from, nor to discern their worth. This leads me to the conclusion, that the math proposed by the OP is not quite adequate for this purpose.
So while I admittedly did not yet dive deep into solving the perceived problem (or analyzing if it is an actual issue), I would tend to change how humans are valued in the metatype column. A first idea would be to keep humans limited to a max of C, but give them slightly more adjustment points than the metas get at each priority.