The LVCVA is a public authority and the Loop contract had substantial penalties in place, about 30% of the total ~$50 mil amount, tied to various tiers of service the system needed to hit. So right off the bat not only do you have good financial incentives for a good process, but strong legal risk and public scrutiny if things go bad. While the test was not public (for these very reasons), it was attended by auditors from BDO LLP - LVCVA's long standing auditor (and not one hired for the occasion by BoringCo).
The accounting company affirmed the test results, CFO Ed Finger told the board that the auditors observed 157 unique rides and there were no negative findings. These are all public records and board minutes anyone can request and consult.
Essentially, according to what was reported in the media, they had the few hundred volunteers board packed Teslas (3 pax + driver), ride the system, disembark and take another ride, while remaining within the station limits.
I wouldn't call the 4400pphpd result an "extrapolation" - it's a real, instantaneous capacity number once the system reaches steady state, just like you don't have to drive for an entire hour to express your speed in km/h or mi/h.
The figure is of course not indicative for the real rush hour capacity of the station infrastructure, especially the underground stations that have escalators etc., when they are packed with disoriented tourists carrying luggage and not necessarily making an effort to move fast and hit good numbers.
Hence, I think my original point stands firm, Boring has demonstrated vehicle rates that can beat light rail, the basic premise of cheap narrow tunnels is sound. To actually demonstrate competitive numbers in real life scenarios will require larger vehicles, better organized ingress flows and procedures, more and perhaps larger stations etc. But these are all tweakable factors depending on actual demand at a specific station, they can run a mix of vehicles and expand infra where it makes financial sense, but only IF the tunnels have good vehicle rates, ie. no more than seconds of headway with extraordinarily rare in-tunnel break downs.