No, a movie of a car is not a car simulation. It ILLUSTRATES the car, so it is a DEPICTION of the car, but there is no substance here. No properties of the car are mapped to any properties of a simulation such that there is a single mathematical/logical MODEL which describes the behavior of both of them. Without that, you simply have a depiction. A SIMULATION must be an ANALOGY. It has to contain some essential features which stand in for essential features of the simulated thing. Without that, its something else.
I think the 'granularity argument' does have some essence of truth in it. That is to say, if a simulation only corresponds with the thing it simulates at a very few points, then it is likely that the ways in which we want it to be an accurate analogy will fail. Still, as you point out, its circumstantial.
Honestly, I'm a bit more pragmatic about it. I don't mind the idea of filling in the blanks, so long as there is some guidance as to what those blanks actually are. For example, if we took a series of still pictures of a car on a road.
1. Car is driving on road. It is raining.
2. Car is going around corner, the tail of the car is way out of alignment and the car is obviously skidding.
3. Car is in the air, upside down, the guard rail around the corner is broken.
4. Car is a ball of fire on the ground.
Is this a simulation of a crash? Maybe not. But, it doesn't take a genius to fill in the blanks here. If we take out pictures 2 and 3, we have no idea what happened. But, with a bit of granularity, we can make some pretty educated guesses. So, as far as I'm concerned, it's close enough to a simulation that I'm not going to get too fussed about the exact nomenclature.
That aside, I do disagree with
@Ovinomancer in that the jumping rules are a simulation of anything. Again, the lack of granularity hurts it. By the rules, a halfling, a human, a horse and an elephant all jump exactly the same distance if they all have the same strength. That's not much of a simulation. That's much more a simplified rule of thumb for an action that doesn't come up all that often but, needs something for when it does. So, we use "Move 10 feet, jump STR in feet." It's simple, it's easy and it gets the job done. But, if that's bar for a simulation, I'd say that's a very, very low one. True, you do get the point across - player declares that his character jumps, character jumps X feet, done. So, it does have a fairly 1:1 correlation between what the player declares and what happens in the fiction, so, there is that. But, as a simulation, it's seriously lacking.