Four Minutes of Trackmania Perfection | [TAS] E02-Endurance No-Cut

As you may know, E02-Endurance is my favourite track. I've been rather obsessed with it for a long time. After a lot of effort, I finally managed to finish my long standing dream of seeing the perfect E02 no-cut run. The current E02-Endurance WR is 4:01.45 by Mawi. The TAS I did here managed to shave nearly 5 seconds off that time, showing that a sub-4 minute run is totally possible (in theory). You may think that this run makes Mawi's WR look bad, but you're wrong: for me, it makes me appreciate how amazing Mawi's WR is even more.

This TAS run was done by combining save-states with donadigo's bruteforce tool. I basically spent 30-60 minutes on each corner trying to get it perfect, and then used 15 minutes or more of bruteforcing to fine-tune it afterwards. If I spent even more time on each corner, more time could be saved. Realistically, I would say that somewhere between 1-2 seconds could be saved still. Perhaps more if you included some tricks, like some potential ramms and speeddrifts. Unfortunately, after getting about 1 minute into the track, I started improving the process that I was using, and I realised that the first 30 seconds wasn't actually that great. It's still much faster than the WR, but it could be better. As the run went on, I got pickier and picker with each turn, so the run only gets better and better.

Some notes:
1. It's not always fastest to drive as close to the wall as possible. You may notice that I don't do this on every turn. It's generally because it's not ideal to do so, for various reasons.
2. There are a lot of ramms in this run. Every one is there because it's actually useful and faster. A lot of these were found by the bruteforce tool.
3. It's difficult to know for sure exactly how to optimise a particular corner. There's always a balance between speed and distance travelled. For instance, in many sections of this track, you have the option of doing 1, 2 or 3 drifts (e.g. at 2:20). Mawi chose to do 3 drifts here. I experimented a little and determined that 2 drifts seemed to be the fastest. But it's hard to know for sure. Of course, one could chose to do 3 drifts because it's more consistent, rather than because it's fastest.
4. The end is kinda a "cut", but not really? I can jump further along the track because I have more speed. If I had enough speed to jump directly into the finish, should I be forced to slow down to avoid jumping over/cutting a small piece of the track? I don't think so, so I allowed this.
5. The replay is publicly available, and can be found here:
https://www.mediafire.com/file/d6aqsv...

Thanks,
JaV