You're glossing over the advantages specific to cryo-EM though, which is it can give you a good picture of large ensembles of small things at this accuracy, wit

You're glossing over the advantages specific to cryo-EM though, which is it can give you a good picture of large ensembles of small things at this accuracy, with a lot less computational/interpretation hardship than X-ray crystallography or NMR. So for seeing structures of large protein complexes and how the super structure varies with heteromeric variation this is a really big deal. If you wanted to see what viral capsids in-situ looked like down to the atoms this is what you'd use. There are a lot of places this is going to be useful, there is no one method to rule them all in structural proteomics simply because nothing can offer all of the spatial and temporal scales one might need. Plus cryo would only give you the surface hull of these complexes, you'd still need a different method to fill in all the exact structures inside. Writing up these methods as if there is some competition for "best" is an entirely false impression to a lay person reading. The best is all methods improve and researchers collaborate to provide every possible scoped view. Even mass spectrometry is doing crazy things for investigating the structure of disordered proteins. You'd never use that for something you would use cryo-EM for.