The cartoon graphics showing mount stupid seem to be exaggerated, popular-scientific representations, and should, as far as I can see, be regarded as schematics to illustrate a more subtle effect.
From what I can find, the cartoons you provide are exaggerated and simplified versions depicting the more pronounced examples of the Dunning-Kruger effect like in Fig. 1 where subjects performed a Wason task (Schlösser et al., 2013). The mountain is somewhat visible here, especially in panel B where the raw scores are plotted. There the subjective performance increases monotonically, while the perceived score in the poorest quartile is above the second, which resembles the peak you are referring to. However, this is one of the most pronounced samples I could find and it may well be caused by variance, as other graphs in that same paper, other sources you provide above (Kruger & Dunning, 1999) don't show this peak. Similarly, plots in Dunning (2011) are less pronounced at this point.
Fig. 1. Actual versus perceived performance on Wason task. Dotted line represents where perceived performance should lay according to the Krajč–Ortmann model. source: (Schlösser et al., 2013)
- Dunning, Advances Exp Soc Psych (2011); 44: 247-96
- Kruger & Dunning, J Personality Social Psych (1999); 77(6): 121-34
- Schlösser et al., J Economic Psych (2013); 39: 85-100