This is an awkward situation. They are the same person. And they are not. So nobody is wrong here.
Both profiles are by the same researcher, who seems to have recycled his data into a new, later profile for the same individual, but wanted to distance her from another, incorrect, iteration of herself with the impossible parent Hugh Courtenay of Haccombe. That Margaret, had she existed, might have been born circa 1417 but could not have married a man who died in 1381.
Either Courtenay-9 should get the father Hugh of Haccombe, with the caveat that if he had such a daughter she was married to neither John Cobham nor Theobald Grenville, or, if she is to granted no parents, her birthdate needs to be altered (as, so far as I can see, it is only based on that presumed parentage) so she can be merged with "Margaret the younger" Courtenay-1057, who was the wife of Theobald, but could not also have been the wife of Cobham (married to her presumed elder sister, another Margaret, Courtenay-67).
Consequently Courtenay-1057 is speculated as a second, younger, daughter named Margaret born to Hugh Courtenay, 10th Earl of Devon. But it is the two latter Margarets who should not be merged.
My thinking is that it is probably preferable not to merge Courtenay-1057 and Courtenay-9, but simply to attach Courtenay-9 to the incorrect parent as she already has a disclaimer of being an "imaginary person".