Test and document willStartLoading & didFinishLoading delegate methods

[friedbunny]

With #630 and #2627, the map loading notifications should be being sent from core MBGL, which means that the mapViewWillStartLoadingMap: & mapViewDidFinishLoadingMap: delegate methods that already exist in MGLMapViewDelegate should be ready for use.

• Make sure the current delegate methods are actually hooked up
• Update core changelog
• Update iOS docs
• Is this testable? If so, write tests.

/cc @incanus @1ec5

[friedbunny]

Going to wait for the ios-improve-tests branch to shake out before thinking about writing tests for this.

[friedbunny]

This doesn't necessarily need to wait for the MapKitization proposed in #2775, but that should be on the radar here.

[1ec5]

What’s left? Should this still be on the 3.0.0 milestone?

[friedbunny]

What’s left? Should this still be on the 3.0.0 milestone?

I see two options:

1. Ditch this and only surface these delegate methods when they're improved and work reliably (Make iOS delegate notifications consistent with MapKit #2775).
2. Merge but add notes about limitations to the inline docs, stating that functionality will improve/change eventually.

I'm leaning towards option two, though either way developers will be able to access these methods.

[incanus]

Just to restate, the issue is that these don't work reliably for tests, or always?

Option two sounds better to me providing these work for users, and we'll continue to up our testing game across the whole framework, though we'd need to update CHANGELOG.md to not be so bright-eyed and bushy-tailed, I think.

[friedbunny]

Just to restate, the issue is that these don't work reliably for tests, or always?

Per #2775, these two callbacks (and the related load failure callbacks) don't work as a MapKit developer would expect — they reliably fire, but at the wrong times.

[incanus]

Ok, I think we can roll this PR in then for 3.0.0.

[incanus]

Rebased & awaiting tests (even if this is just header doc changes).