The first case is well documented. 'Nuff said.
The second instance isn't built on a single case but rests firmly on the assumption that fomites easily spread the virus. You are infected, sneeze on a doorknob, and anybody who comes along and touches the doorknob has a good risk of contracting the disease. This is a wonderful idea and allows for a lot of caution and protective practices.
Sadly, there's no evidence that the virus spreads much that way. It's a nice-sounding assumption, but that claim to fame rests just about entirely on sounding nice and on its being an assumption. It's a good thing for scaring people, so there's that.
The third instance, early on, looks solid and at the time led to the conclusion that's given. That was before it was known how easily asymptomatic "sufferers" could spread the disease and before it was known how many asymptomatics there were. There could have been the one spreader on the bus. There could have been one or two others. A few days later when the spreader of notoriety was identified there'd be no way, at the time at least, of identifying others as 2 days into their infection when they road the bus.
In other words, that example's been undermined, but unless you stop and think about what was known then (and built into the analysis) and what's known now (that shreds some of what was built into the analysis) you'd miss that. Epidemiologists have been things to do than explain why previous findings are wrong, esp. if those previous findings still encourage people to do wise things.
Often we actually know less than we think.