Optionally send events asynchronously back to the user

[mlandauer]

These events should include things like:

• Compile started
• Log output
• Compile finished
• Run started
• Run finished

On the start of the run the user provides a callback url with optional headers that can be used for authentication

[mlandauer]

For the callback url and header information we have two options. In both cases the url and headers are provided when the scraper is started but then either:

• The values are stored somewhere that only the clay server has access to and every time an event is created the server looks up the callback information.
• Or the url and header information is passed to the clay server for each event.

The first option is better from a security point of view because in the second setup the scraper author could send fake events to the clay server which could specify arbitrary urls. It's much better to trust the clay user (who creates the run) where to send the callbacks.