The fresh timesaves over my record from 8 months ago include:

• Allowing Maxwell to move/shoot during tutorial textboxes by making him start auto-pathfinding before a trigger (I already was doing this for the plane in 0-4, although there are newer and easier applications that I was made aware of). New applications found by JSRphones when doing a tool-assisted demo.
• Getting 35 extra ollars in one of the playthroughs of A1-1 by summoning a Living Room (this multi-object doesn't make the par go up, so you don't lose any money by summoning the unnecessary object, and you get the All New, New Object, and Entertainer merits. Overlapped with the Genius merit to mitigate the timeloss of the merit display). This is a small amount of money, but without it I always had to play A1-1 one additional time, since I would land just short of 4000 Ollars.
• Doing the Action levels in Mish-Mash at the end of the run, instead of puzzle. Before Tank clip was used in runs, it wouldn't have made much sense to do Action levels since A10-1 would have taken too long, and it's counterintuitive to be forced to play A10-11 since it is slower than P10-11. However, the first three levels you play save plenty of seconds optimally, and with my reroute to do Action levels in Peaks, it makes sense to stay on the same page for the whole run. (Even if we were still doing Puzzle levels for Peaks, it would still be faster to do Action levels in Mish-Mash.)