Why does it take so much effort to use a garbage collected runtime efficiently? Wasn't the purpose of a GC in the first place to make things easier for the developer?

It actually does remove a lot of hassle from developers. But with every abstraction, the complexity is not actually removed, just hidden. This means that in situations where the abstraction fails, you would need to still understand the hidden complexity ...

Tuning a few knobs on the GC is still a heck of a lot easier than managing memory yourself.