I've seen these splits quoted before, but they are quite a bit out!
Warren went through 400m in 54.55 (factual evidence on any recording of the race), and Ovett was at least 3m back in about 5th place. From my copy of the race (backed by splits given in Athletics Weekly at time) I'd say he (Ovett) was no faster than 54.9, giving him a 2nd lap of 50.5, as his winning time was 1:45.40
Coe finished in 1:45.85, meaning he finished 0.45 behind Ovett. If the above splits were correct, he'd have to have been level with Ovett at the bell. He clearly was last at 400m, another 2 or 3m behind Ovett. His actual 400 split was more like 55.2. So his last 400m was c. 50.65; much faster than the 51.2 quoted above.
The last 200m of Ovett certainly wasn't 24.3!! He covered the last 100m in 12.7 (Coe 12.6), so he'd have had to run the penultimate 100m in 11.6 to make the 24.3 true. He didn't. His last 200m was 25.0, Coe's 24.9.
Ovett covered most of the last lap on the inside of lane 1, whereas Coe ran both last bends in lane 2. So he ran more like 407m in 50.65, which works out at about 49.7/49.8 had he ran only 400 and not have to run on the outside.
Some of the last 400m splits above are slightly out; some being quoted by commentators at the time (like Cram's 49.9) based on what their statistician saw the clock was (roughly) at the bell. On close scrutiny of Cram's Europa Cup race in 1987, he actually covered the last 400m in 50.1. Gonzales was practically alongside him at the bell, and beat him by a few hundredths, so if anything was slightly faster; perhaps 50.0.
I get slightly different time for: -
Morceli 91 WC- 51.6 (3:32.85);
Cacho in 92 I have at 50.5 (3:40.12);
EL G 2004 - 52.0 (3:34.2)
Other notably fast finishes were: -
Bile 1987 WC - 51.5 (& last 800m in 1:46.9) in a 3:36.8;
Cram (50.9) & Coe (51.0) in the European Champs in 86 - 3:41.09/3:41.58;
El G - 53.9 in the 99 WC (3:27.65)
The thing about the time for the last 400m is that it doesn't tell the full story, unless you also know say the last 200m or last 800m.
If you look at Coe's 84 Olympic win in 3:32 and compare it to Morceli's slightly slower 91 WC run, it would appear, looking only at the last 400m (51.6 Morceli to Coe's 53.1) that Morceli was finishing quite a bit faster. However if you look at it in the context of their last 800m (1:49.9 Coe, 1:50.1 Morceli), last 200m ( 25.7 Coe, 26.0 Morceli) and last 100m (Coe 12.7, Morceli 13.3), it's clear that both were very close in the final stages, just that their finishing speed was distributed differently.
Coe's last 200m was 24.1 (last 100m in 11.9) but the finishing time was 1:47.28 not 1:46.3.
Equally impressive was his last 200m in 24.7 in his 1:44.50 European Champs win in 86, where he also ran some of the last bend wide. Probably worth 24.5 with a run on the rails.
I'm not sure anyone has ever seen Aouita's supposed 36 last 300m run!? At the time it was reported no final 400m time was given, so I doubt that last 300 split was legit.
Aouita covered the last 400m in 52.9 (Cram in 53.4) in the WR race in Nice 85, NOT 51.7!
EL G's last 400m in his 3:26.00 WR was 53.5 not 53.1 according to The "Progression of IAAF World Records 2007 Edition" (published by the IAAF - though they're stats have some mistakes) and I got the same from my copy of the race.
I guess it depends a lot on where people read these splits (many of which are not official, of course). I have always found the most reliable ones are found by just playing the races and checking them out oneself. Then of course that can lead to it's own problems. Some tracks have clear marks for the 100m from home, while other don't, instead having one at 110m and another at 90m!