I think what they're talking there was maintaining a bank of small samples taken from the populace over time so that chemical tests and other research can easily be retroactively performed. That whole section is about monitoring PFA levels in the general populace.

I don't think an actual human milk bank for consumer use would be physically possible, volume wise they would need to be milking a bunch of women like actual dairy cattle. And logistics wise the spoil time on breast milk for newborns is measured in hours at room temp and only 1 week to 1 month when frozen.