As I said, the evidence I provided is anecdotal. This means I can't say it proves anything, but it also means you can't dismiss it with certainty.
Now if we continue on this anecdotal path, did you ever go to Markarth during your hundreds of hours of gameplay? I verified the CPU bottleneck by lowering resolution from 1080p to 1600x900 - there was absolutely no change in framerate. Most of the time framerates were GPU limited but there were situations like Markarth where it was not the case.
The link you posted is irrelevant to the discussion of whether a 3.2GHz first gen i7 is a bottle neck. The 4.2Ghz i7 is a good 30% faster. Regardless, the link is relevant only to a limited number of games, and those games are only tested in the limited benchmark scenario which may or may not be representative of actual gameplay. And the average framerates can conceal differences in minimum framerates i.e. framerate stability. Anecdotal evidence: in Max Payne 3, I saw very little increase in averages by upgrading to 3770K @4.2, but I created framerate over time data which showed improved framerate stability.
I upgraded my CPU because I had real reason to. Almost none of the games I played ran the way I wanted with the i7-920 @ 3.36 (couldn't clock it higher than that), and the 3770K fixed that. Mostly. Planetside 2 is still bottlenecked even at 4.7GHz, that's just cos it's poorly optimized.