I checked it and it looks right to me.
Makes sense, since the Lambert cylindrical equal-area projection has worse resolution-efficiency than the central cylindrical projection, given the same cutoff. And both are worse than Mercator.
However, if you choose the standard parallel that optimizes resolution-efficiency, instead of setting the standard parallel at the equator like Lambert does, then the optimal cylindrical equal-area projection has the same resolution-efficiency as the optimal central cylindrical projection (noting that the optimal standard parallel is different for each case - for the central cylindrical, the equator actually is optimal).
Probably an unintentional side effect of the recent forum update? Older posts still show the image, but new ones only show the link.