Thanks for the links and taking the time to research this. The trouble is with these world checklists of plants, whether it be CoL, IPNI, The Plant List, Tropicos at Missouri Bot Garden, Kew, Royal Botanic Garden, none are in full agreement and consensus, and there are often lots of mistakes in them.
Example: CoL does not recognize Epimedium × omeiense Stearn 1995 (natural hybrids between E. acuminatum x fangii), so cultivars of this botanical entity will be missing from CoL.
CoL does recognize Epimedium sasaki Maekawa 1955 (as if a species!), however it was published as a name for natural hybrids occurring in Japan, as E. x sasaki Maekawa 1955, so perhaps CoL did not interpret this one correctly.
For Epimedium, I reference the Father of Epimedium and Berberidaceae, William T. Stearn and his monograph The Genus Epimedium and Other Herbaceous Berberidaceae, 2002. I also depend on the work by Darrell Probst who spent decades on Epimedium taxonomy and botanical exploration in China, Korea, and Japan; he also collaborated with W.T.Stearn.
The name E. versicolor was published in 1854.
The name E. x versicolor was published in 1849 (mentioned in the Epimedium book). In the early days of taxonomy (1800s), yes indeed some groups of hybrids were given latinized names to describe the crosses, whether they were man-made or natural.
So, how best to interpret? If the NGA prefers to strictly adhere to a singular reference, so be it. For me personally, I'm following the taxonomy as developed by those who worked decades in the field, and always try to find consensus among the most reputable taxonomic authority for specific genera, there are none that authorizes on all genera.