Apparently, the diagram on Wikipedia page is not correct (either outdated or oversimplified). According to what I could google, the research does show a hump in L-cones sensitivity at violet colors, for example:
The CIE color matching functions are a very interesting topic, and a subtle distinction between cone response and perceived ability to match a given color. The main issue being that we haven’t had any ways until recently to directly measure cone response. The color matching functions are derived from clever experiments that attempt to figure out cone response based on what humans think they see, without opening their heads.
https://www.yorku.ca/eye/specsens.htm
https://www.researchgate.net/figure/Normalized-spectral-sens...