Meant to ask Charlie, do you just slowly zoom and pan as the shots are being taken? Trying to get my head around what you have done here
2 separate timelapses going at the same time, A9 @21mm, A7r3 @50mm? It's better if I had 2 @A7r3's but I'm ok either way, cant live without the A9 at this point 
Timelapses should be overshot by quite a bit for easier post work, 1200+ frames each camera, 5 second interval, AV mode, slightly high ISO to make it work.
Process both timelapses to get as close to each other as possible, sort of hard matching them especially with heavy cropping involved.
stack the layers in after effects, keyframe transitions and scale as needed. null object to tie everything together, new comp with 1080p as final output. The rest is matching up music and zoom effects all done in post. So the 10s mark, theres a transition from 21mm already cropped to 50mm and then zoomed all the way to 150mm fov or so (R3 can crop a lot). pan around with some key frames, zoom out, 3/4 of the way, you hear a drop beat and quick movement, that's the transistion from 50mm back to 21mm of the other camera (this transition is a tad sloppy, but I dont know how to go redo this section without dumping it a lot of it). Lots of trickery 
I have a lot to learn on the time lapse side it seems
