My default assumption is that when there are claims of harassment, there is some substance to them, even if the specifics are wrong. Bringing forward claims of harassment in the tech world turns your life (and that of the accused, if their name is publicized) into a shitstorm; people don't tend to do that without reason. But this is just a heuristic. It looks like Horvath's claims of harassment might be unsubstantiated in this case.
It's hard to tell, though, because most of the things she complains about (with the exception of the code-review stuff, which appears to be just false, which is worrying) is unverifiable outside of he-said/she-said. Note the noncommittal wording: "The investigation found no information to support misconduct or opportunistic behavior by the engineer against Julie or any other female employees in the workplace." No information to support suggests "neither evidence for or against". Which is natural. Harassment is hard to prove or disprove precisely because it usually doesn't leave evidence.
> there was an apparently thorough investigation by a reputed external investigator who found that the claims were false. [...] And yet this single person's claim was enough to force out the CEO
Specifically regarding the ousting of the CEO, let me quote the investigation you linked:
> The investigation found Tom Preston-Werner in his capacity as GitHub’s CEO acted inappropriately, including confrontational conduct, disregard of workplace complaints, insensitivity to the impact of his spouse's presence in the workplace, and failure to enforce an agreement that his spouse should not work in the office.