I think VT49 (ca. 1965) is carelessness. PE22:164 shows that neither the concept of an inherently durative verb nor the roster of those have changed during the decades. Of course, these pieces of evidence might be used as an equally strong proof that Tolkien specifically eliminated the particular MEL out of this class — but I believe it improbable. EtyAC N melo which never got a chance of being obsoleted, also supports here.

I won't be afraid to use both, since both are attested.

And what about S? In VT/45:34 we have a deleted N melant, though we see S úvel (supposedly from *mel-) in PE/17:144-45.