There's a pretty bad bias problem here because the probe app scans domains that are frequently blocked in dictatorships, but doesn't scan domains that are frequently blocked in "democracies", like Anna's Archive. Then it ends up looking like the dictatorship countries have all the censorship.
Internet blocking is a last resort for regimes that can't do anything against the source of the information or intermediaries, but it's a red herring in regimes that can. Anna's Archive has a political position and is one of the very few voices that survived the immense pressure. How many didn't? How many attempts were not performed at all because of it? This suppression is really hard to measure, OONI can't do anything here, it measures blocking. It's very simple to understand.
I would only use the term "censorship" to refer to the prevention of speech based on its content. Content-neutral time, place, and manner restrictions aren't censorship according to the way I use the word, and neither is copyright protection.
Note that censorship isn't necessarily bad and restrictions on speech that aren't censorship aren't necessarily acceptable. The prohibition on child pornography is censorship but I support it, while I am opposed to IP law in its entirety.
It all depends on what you're censoring.
Is it a bad sign if CSAM, Personal (medical, financial, etc) Records, Snuff, and Malware/Ransomeware are censored? What about manufacture of WMDs, terrorism, hate speech? Free software, and copyrighted music? Historical records of genocide?
Scare quoting democracies is probably telling, but OONI absolutely does test content blocked in Western countries. https://explorer.ooni.org/domain/annas-archive.org