The line is long enough for the curvature of the earth, and the subsequent distortion in the map projection, to be relevant. Notice that they're closer to a straight line on Mercator than to the geodesic: that means if you were to use purely local referencing to align the sites, they wouldn't end up where they are. You only get them to line up when you distort the natural geography with a projection of some sort, so if you want to make an argument that they were intentionally built on a line, you also have to account for the systematic deviation from the geodesic. And that prompts the question of whether that's remotely feasible given what we know of the history of cartography.

What I'd want to know is how old the story of St Michael's Sword actually is. Not the churches, but what's the earliest reference to them being in a line. My bet is that it's well after Mercator, and probably safely after the 18th century, when chunks of Europe got geodetic surveys done.