I think that, in general, a TAS should aim to be better than RTA runners in all parts of the game, but how much better is really up to the author. What I mean is that I know a lot of TASers (myself included) get frustrated when they spend a lot of time on one small part of the game so they can save a fraction of a second. In most cases, I think it's better to do the best you can within a reasonable amount of time and then move on to the next part of the game. It's just about balancing how much time you spend working on something vs. how good the final product is. In other words, you get diminishing returns when it comes to optimizing a TAS.
That's not specifically a problem with movie rules, it's just something I've noticed about TASers over the years.
I can give you a good example of what I mean. For a little while, I was working on TASing the Wii version of Call of Duty 4, and I think I got
really good at the movement. I won't go into all the specifics on how the game works, but the movement in that game is similar to how it is in Quake 3: Arena. So the movement for most of the game revolves around bunny hopping and air strafing, both of which are very complicated to optimize. This is made worse because Dolphin doesn't have good TAS tools compared to BizHawk or libTAS, and even the custom versions that have LUA scripting don't support the Wii motion controls this game requires. So all the inputs have to be done manually. The Wii version also runs at a lower framerate, which consistently loses time over speedrunners that play on other versions due to slower movement. However, even with all that, Wii TAS movement can beat PC speedrunners over short distances (but loses time over long distances for reasons I won't get into). The only problem is that air strafing is very annoying to optimize manually, and there's currently no way to do it automatically. That video I linked before took me days to make, and the F.N.G. training mission I TASed a few years ago took me over a month, for just 18 seconds of gameplay, and it wasn't even close to being optimized.
Anyway, what I'm getting at is that an optimized TAS for Call of Duty 4 on Wii just isn't going to happen. A perfect strafe jump on Wii can get you up to 385 speed before you hit the ground, which is pretty good even for a player on PC. If I were going to TAS the Wii version, I wouldn't spend all the time trying to get that perfect amount of speed every time. Instead, I'd go for about 370 per jump, which is considered average for a PC speedrunner. Still, that's 96% as fast, which is not noticeable for any casual viewer, but the difference is that a good TAS with slightly slower movement could be made in a fraction of the time. A TAS where I run across the airstrip could be made in one sitting, and not over the course of several days. That training mission could have been made in a few days instead of over a month, and it would probably only be a fraction of a second slower.
One small side note: A Wii TAS would probably be slower than the PC world record, but it's because there are a lot of version differences. So it being slower than a PC speedrun wouldn't necessarily mean the TAS is unoptimized. For other games where there aren't any major version differences, TASes should still strive to be faster than RTA runs in all parts of the game, but I don't believe they have to be perfect to have a good place on this site.
I know I'm getting off topic because this thread is about movie rules, but I wanted to clear up what I meant about balancing the amount of time you spend making a TAS vs. how optimized a TAS needs to be.