Keep in mind that in addition to the "learning curve" arguments, there is also the functional developer ergonomics of things like live reload (where you can maintain memory state) that are simply not possible to do with C++ without heavy limitations or customized tooling.
Being able to fix a bug without resetting memory state is a huge ergonomic advantage in game development where generating the right memory state can be incredibly complex and depend on a ton of very specific and hard-to-reproduce factors. Not to mention recompiling and restarting a game can be incredibly slow.
Yes, but it has many known limitations and isn't nearly as reliable as a runtime that has a full GC and virtualization optionality, which you really need in order to fully track what state can be evicted and what needs to stay.
Being able to fix a bug without resetting memory state is a huge ergonomic advantage in game development where generating the right memory state can be incredibly complex and depend on a ton of very specific and hard-to-reproduce factors. Not to mention recompiling and restarting a game can be incredibly slow.