Raise informative error on batched resolver timeout

[jordi-chacon]

When the Task that runs a batched resolver times out we currently raise a generic error like this:

** (exit) exited in: Task.await(%Task{owner: #PID<0.15353.21>, pid: #PID<0.15380.21>, ref: #Reference<0.2441152191.2237136897.192020>}, 5000)

This provides no context to the engineer debugging the error about the resolver that timed out, making debugging much harder.

This change raises a RuntimeError which includes the resolver that timed out along with the arguments it received:

Batch resolver timed out after 1 ms.
Batch fun: {Absinthe.Middleware.BatchTest.Schema, :slow_field_by_user_id}
Batch data: [3, 2, 1]
Batch opts: [timeout: 1]

This should make the debugging experience a lot better.

[benwilson512]

This is a great change, can you add a changelog entry?

[benwilson512]

I think the main thing I need to sort out is whether turning this exit into a raise is going to be an issue.

Task.await/2 emits an exit when it times out, instead of a RuntimeError that raise emits.

@maartenvanvliet can you think of any reason why this could be a problem?

[rossvz]

Bumping an old PR to see what it would take to get this merged. We see hundreds of these errors and we use batch resolvers everywhere so its hard to track down!

If there's hesitations around raise vs emitting exit - could the code be tweaked to Logger.error or something similar?