Just tried on SSRX with the following setup:
any car with a setup so you dont vary and variable speed at a certian point (300km/h at the pit exit).
first run: go full speed and hit brakes, see distance,
second run: go full speed and both hit brakes and downshift to lowest gear, see distance is equal.
It seems the game does not have engine brake at all.
Just tried a the 2022 ND MX-5, bone stock.
SSRX, from steady 100mph, (limited by a wedge under the pedal), in 5th. Timings taken from the replay.
Coasting in neutral (Shift straight into Neutral)
At 100mph - time taken to coast to a standstill: 201.372s
Coasting in gear (Shift down one gear when the rev bar drops off the left until 1st gear)
At 100mph - time taken to coast to a standstill: 79.217s
Engine breaking very much does exist in the game. In the test you conducted, how much the engine breaking affects the stopping performance relies on the differential between engine braking and friction braking being noticeable, that's a lot of variables. I thought I'd try something similar anyway.
Viper Gr.3, bone stock.
SSRX, from steady 140mph, (limited by a wedge under the pedal, in 5th). Timings taken from the replay.
Braking in 5th (leave in 5th to standstill)
At 140mph - time taken to reach standstill: 3.800s
Braking in gear (shifting straight down the box to first)
At 140mph - time taken to reach standstill: 3.716s
So, marginally faster, since I'm going from the times in the replay, I'd guess there'll be a little variance, but I bet if you did it 10 times in a row, at least 9 of them would be quicker. Also worth considering that the engine braking acts on the rear wheels in the Viper, which offer less retarding traction anyway. But in any case 0.084s in each big stopping zone could be the difference between Gold and Silver.
In both cases its also worth pointing out that by the time you reach first and the revs are dropping anyway, the engine braking effect is much less, it all happens at high RPM.
To
@Gio318is 's point, the revs do indeed fall faster when the clutch is disengaged, and slower when the clutch is engaged.
Again, with the clutch engaged the rotational mass trying to turn the engine is higher, it's like having a much heavier flywheel. I'm not saying it's 100% accurate, and I'd be surprised if it's accurately modelled based on real world weights, I'd imagine more just a fixed factor or percentage, but there's an intuitive element to it at least. I'm curious to see what my car would do in the real world, but since it's still having its bottom end taken apart, I can't.