Fry relies on McGann, so in this answer I’m going to summarize McGann’s argument.
McGann starts with the observation that a commonly accepted editorial principle is that the editor should try to recover the text that comes closest to the author’s final intention. In general, later drafts are preferable to earlier drafts, and publications corrected by the author are preferable to uncorrected publications, and to publications edited by third parties.
[The Indicator] text is the only one Keats ever published, for he did not choose to print the poem in his 1820 collection […] The poem first entered Keats’ collected works in 1848, in Richard Monckton Milnes’ Life, Letters, and Literary Remains, of John Keats. In this 1848 printing, however, the text was taken from a copy of the poem made by Charles Brown. No one knows the source of Brown’s text, though editors have conjectured, not implausibly, that it was made from a (now lost) holograph fair copy. This 1848 text, derived from Brown’s copy, is the one we all now read, and it differs greatly from the text printed by Keats himself. The new Harvard edition also prints the Brown/1848 text.
Under the circumstances, the prima facie bibliographic facts would normally, and without any question, demand that an editor—especially an editor in 1978—print The Indicator text, not the Brown/1848 text, for it is The Indicator text which, so far as we can tell, most closely corresponds to the author’s final, active intentions. The question then naturally arises: why has the post-authorial critical tradition from 1848 to the present normally printed, read, and studied the poem in the Brown/1848 text? For a study of that tradition, particularly during the past fifty years or so, shows that editors and commentators are aware of the problem and have all along deliberately chosen not to print The Indicator text.
Jerome McGann (1979). ‘Keats and the Historical Method in Literary Criticism’. In MLN 94:5, pp. 1000–1001.
The explanation, McGann suggests, is that editors and scholars have preferred the Brown/1848 text over the Indicator text, and this preference is strong enough to overrule the accepted editorial principle. If this is right, it is worth looking at the differences between the two versions to see if they suggest a reason for this preference.
So I think Fry’s summary of McGann, quoted in the question, is misleading if not outright tendentious. McGann didn’t start with the (very minor) differences between the two versions and then try to inflate them into a judgment about which poem is more politically correct. Instead, he started with the observation that editors like Jack Stillinger (editor of the 1978 Harvard edition) had gone against their own principles in publishing the Brown/1848 version, and then tried to determine the reasons for this by looking at the differences.
Coming at the question this way round, it is noticeable that
the result of the Brown/1848 changes was to make the character of the knight more sympathetic to the reader and the character of the elfin lady less so.
McGann, p. 1002.
||And there she wept and sighed full sore,
And there I shut her wild, wild eyes
With kisses four
|And there she gaz’d and sighed deep,
And there I shut her wild sad eyes
So kiss'd to sleep.
||And there she lullèd me asleep
||And there we slumber’d on the moss
Although the differences are slight, passages (2) and (3) express more of a mutual love-making in the Indicator version, and more of a seduction in the Brown/1848 version. The reader of the Indicator text is more likely to interpret the wretched wight as a lovelorn man who can never recapture his one night of bliss with his fairy lover, and the reader of the Brown/1848 text is more likely to interpret the knight-at-arms as the victim of a magical enchantment.
After 1848 readers of Keats characteristically saw the elfin lady as a sort of demon lover who had ensnared the unsuspecting knight. […] A persuasive case could be made that Brown and Milnes share the immediate responsibility—not merely in fact, but by ideological design—for giving us the idea that the elfin lady is simply demonic, a sort of uncomplicated lamia.
McGann, p. 1003.