JS: Expose whether an endpoint name is synthetic

[asgerf]

The EndpointNaming library was written to support two use-cases: the model editor, and VEA.

However, I think they differ in how they should deal with the "semi-internal class" problem:

class InternalClass {
  m() {} // exposed to client code, but not reachable via an access path
}
export function get() {
  return new InternalClass();
}

For the model editor use-case, it seemed best to synthesise a name for InternalClass.m() above, rather than give up entirely.

This is not ideal for the other use-case however, so this change exposes a boolean parameter to know whether the name is synthetic or not.

[RasmusWL]

Overall looks OK to me. (but all this code is new to me)

I wonder if fallbackModuleName should also be considered a synthetic name? Currently it's only used for classHasFallbackName, so it doesn't have any immediate impact, but I'm just wondering if we could end up in a situation where using fallbackModuleName only gives 50 in penalty, so we actually synthetize a name without treating it as synthezied?