As @Matt has said, there is no way (at least short of NSA or NASA level software) to separate multiple exposures on film.
In your case, it's worse than that. Kodachrome, as a slide film, loses information when overexposed (much the way negative film loses information when underexposed), and 3x exposure (about +1.5 stops total) is well out of the exposure latitude for Kodachrome. It's as if you'd metered that frame for ASA 8 instead of 25. Especially the lighter areas of the frame (sunlit portions of the ash cloud, for instance) likely contain no dye at all, no information that could be teased apart.
Even in darker parts of the frame, where the image isn't completely burned white, there is no reference point to separate the three overlaid images. The process you're after is called "deconvolution", but doing it in a way that produces something close to the original image is like correcting the First Light images from Hubble, before astronauts installed the optics to correct the hyperbolic primary mirror -- it requires having a pretty close idea of the aberration that damaged the image in the first place. In the case of a multiple exposure, that isn't possible. All you'd wind up doing is using the triple exposure as a guide to create a pixel-by-pixel hand-drawn digital image -- the task of at least hundreds of hours.
Take comfort in the fact that you have six more frames of the St. Helens eruption than I do (I was three hundred miles away in Idaho at the time), six more than almost every other photographer on Earth (not even counting those who weren't born yet).