Finally got analysis of requested videos done, sorry for delay, busy days and could only give it little time.
I didn’t analyze all videos because some had issues, but some good results are available still.
1999 Schumacher, 2003 Raikkonen and 2009 Vettel have, as pointed out, commentary.
Vettel’s isn’t actually that bad, large parts of lap are clean while for the others possibly with some filtering/manual limitation of rpm range to search in, I could get something, but would take time and then wouldn’t be accurate so, also considering another 2009 lap, just bit slower, was available anyway, I didn’t bother. On MS’ and KR’s laps on the contrary is quite invasive making result just too noisy unfortunately.
On 2003 MS and 2013 Webber’s presumably something went wrong in the editing/compression (probably some frames got lost or something like that) as the lap duration measured on the video is shorter, respectively by about half second and 1s, than should be, which is more than enough to affect the travelled distance, relative position of corners’ minimum speed etc. and make comparison with other laps, and lateral acceleration, invalid.
Luckily I had another video of Webber’s lap downloaded last year, worse image quality but good audio and right duration so analyzed that one, for MS’ unfortunately I don’t have a backup.
For a couple of others then I could only do a partial analysis.
For 2001 MS I only did the rpm estimate:
Reason is I have no speed reference data, and the track was different (I don’t have a scaled image for the old layout) hence different was the racing line and its total length, so I can’t convert the normalized speed, relative to peak, to an absolute value, at least not with acceptable accuracy.
If you have sources/data to solve that, just post them and I’ll complete the work on it.
Finally the 2010 Hamilton video, as you can see it’s not the complete lap, it stops at 130R and in particular misses the speed trap after it. I analyzed it still and used as reference the speed data for the intermediates, but, also considering the unknown on exact travelled distance, the accuracy in speed estimate is worse than it could be so probably just good as a more qualitative comparison:
Too bad as the rpm signal on the contrary was very clean, all downshifts very neatly distinguishable etc. it was a perfect sample, just not complete.
That leaves basically only 4 usable laps of different years, and here speed vs position on track (keep in mind 2004 was damp track so it’s hardly representative of car’s potential, the fastest lap of race was 1” quicker for example):
The time gap is always 2014’s vs the relevant lap.
Artur Craft wrote:
[...]
Top speed trap in 2004 QLF : Brazil-324kmh , Bahrain 324kmh, Barcelona 326kmh, Canada 341kmh, Hockenheim 332kmh, Monza 363kmh
Top speed trap in 2014 QLF : Brazil-343kmh , Bahrain 329kmh, Barcelona 338kmh, Canada also 338kmh, Hockenheim 332kmh, Monza 354kmh
[...]
Remember that the speed trap is usually located in DRS areas, so even if in these specific points the 2014 cars can match/beat 2004’s straight line performance, it doesn’t mean that same can be done in all other accelerations, without DRS (plus there’s the possibility that maybe there isn’t enough energy to use full MGUK power whenever needed, albeit that will probably apply only to longest tracks, at least for Mercedes).
This year DRS was way more effective than in V8’s years, as rpm limit is so much higher than peak used, that in practice there’s no rpm limit, leaving all margin needed to get to the power limited speed; previously DRS speed gain was rpm limited, not power. (on top of that, the opening gap was also increased, allowing to adopt a longer flap’s chord while still being able to “flatten” it)
won't delete, though, as I will, then, ask if(take whatever time you need)can you compare some Suzuka qualifying laps? The list is this:
=D> 
