VISA only has an average txn capacity of 2,000 tps but their network can handle a peak traffic of 24,000 tps. Nobody designs a system with a specific limit and then assumes throughput will be equal to that upper limit.
That is a very valuable observation. An 'adaptive' block size limit would set a limit of some multiple of the observed transaction rate, but most of its advocates (including me) haven't bothered to look up what the factor ought to be. what you looked up above presents real live information from a functioning payment system.
The lesson being that an acceptable peak txn rate for a working payment network is about 12x its average txn rate.
Which, in our case with average blocks being around 300 KB, means we ought to have maximum block sizes in the 3600KB range.
And that those of us advocating a self-adjusting block size limit ought to be thinking in terms of 12x the observed utilization, not 3x the observed utilization.