feat(ngInclude): emit $includeContentError when HTTP request fails

[caitp]

This adds a scope event notification when a template fails to load.

This can have performance implications, and unfortunately cannot at this moment be terminated with preventDefault(). But it's nice to be notified when problems occur!

Closes #5803

[IgorMinar]

lgtm

[shahata]

wouldn't it make sense to have attr.onerror here as well, even if only for the sake of symmetry with the existing attr.onload?

[shahata]

Also, do we really want to report an error even if thisChangeId !== changeCounter? This is also not symmetric with the current successful load behavior...

[caitp]

@shahata re: the onerror thing, I'm not totally opposed to it, but I think it should wait until people ask for it at the very earliest. Even this event is probably a bit much, but it's been asked for a bit.

As for emitting the error even if it wasn't for the same change counter, I'm not sure it really makes a huge difference, because in practice people aren't changing the expression and running a digest 50 times before the response arrives. But it's a small change to make it work like that, so I guess that's fine.

[cdaringe]

Thanks cait. Looking forward to this merge

[caitp]

Yeah, this should be pretty trivial, @petebacondarwin or @matsko can you give this a quick review and see if @shahata's concern needs to be addressed? I'll talk to Igor about merging this on Thursday

[petebacondarwin]

@caitp I agree with you that we should wait on adding onerror until it is requested. It will be a small simple addition later.
@shahata - I agree with you that we should not bother emitting the error if the change counters do not match since the fact that they don't match means that this response is out of date and so it could be that a subsequent request has succeeded and everything is OK, so emitting an error at that point would be misleading.

[petebacondarwin]

Other than that LGTM

[IgorMinar]

yeah, we should emit the error only if the error is for the current request

[caitp]

PTAL, I think that should be everything

[btford]

Landed as e4419da.