Has it been proven not effective at all points?

No, it hasn’t. You’re linking to a famously anonymous site which doesn’t control for quality because the purpose is to give people a number which looks positive secure in the knowledge that very few readers will notice how the studies showing positive effects are flawed or limited.

There are those who claim that you can accurately predict project timelines from the top down -- high-quality estimates, so to speak.

However, this doesn't work. Talk to anyone in software for any significant duration.

You make accurate high-level estimates, using a multitude of small, low-quality estimates.

I trust many independent peer-reviewed studies (102) plus a large number of other studies (another 49), whose outcomes are consistent with all other studies (efficacy declines linearly from 0-7 days after onset). I do not trust a single, large study focussed after the efficacy window has expired.

The big question is -- why would you trust such a study, if you surveyed and understood the efficacy decline consistent with all other studies?