Add new PR categories for Docs, Spark QA, and EC2

[nchammas]

I suggest adding the following PR categories (with examples of each):

• Docs
  • [Docs] Run tests like in contributing guide apache/spark#2149
  • [Docs] SQL doc formatting and typo fixes apache/spark#2201
• Spark QA / Jenkins / Build
  • [SPARK-2912] [Spark QA] Include commit hash in Spark QA messages apache/spark#1816
  • [SPARK-3076] [Jenkins] catch & report test timeouts apache/spark#1974
  • [Spark QA] only check code files for new classes apache/spark#2184
• EC2
  • [SPARK-2065] [EC2] give launched instances names apache/spark#1043
  • [EC2] name ec2 instances and security groups consistently apache/spark#1344

[nchammas]

Actually, it just occurred to me that perhaps the PR categories should be derived from the list of components in the Spark JIRA project.

[JoshRosen]

The backend is now collecting the information needed to automatically categorize PRs based on the paths of the files that they change.

One tricky problem: how do we display PRs that touch multiple components? For example, pull requests that add Python bindings to SparkSQL or MLlib might touch files inside of /python.

[nchammas]

Sounds good. I'd imagine displaying the PR under each category it applies to is a good way to go.

In the MLlib + Python example, an MLlib reviewer might focus on making sure the Python API for a given method matches the same APIs for other languages, while a Python reviewer might just focus on Python style and best practices.

[pwendell]

I agree with the proposal by @nchammas - let's show it under multiple categories for now. I do wonder how many will overlap with spark core though... we'll have to see.

[JoshRosen]

Alright, pushed a version of this live. It should be really easy to modify the regexes used to specify components, so PRs to refine them are welcome.