Revisit emiting events.

I know a similar issue was closed a year ago.
I was hoping it's time to revisit the possibility of Grunt emitting events.

For my particular use case (programmatically load and run gruntfiles) , I would like to see the following two to start with:
- `grunt:end`: Grunt has drained the queue, and there are no more tasks to run.
- `grunt:error`: Something blew up.

Start by tracing Grunt's programmatic load-and-run path and its existing task lifecycle or event APIs. The issue's proposed outcome is to emit grunt:end after the queue is drained and grunt:error when something fails; inspect the related architecture before defining the event behavior.

Written by the indexing model from the issue text.