Ask user to provide more details if it's a low-effort question

Points to keep in mind when implemeting this:

- Should be triggered on keyphrases like "can somebody help" "can someone help" "anyone help" "help" "help I'm being eaten by a shark".
- Make sure that the message length is lower a particular limit (I highly doubt you can fit a well-formed question in under 100 chars, so that'd be a good limit).
- The response should contain something along the lines of "if you've already described the issue, apologies; ignore me" to account for help requests broken across multiple messages.

No file, test, or entry point is named. Start by locating where incoming messages are matched and responses are generated, then verify the keyphrases, the short-message limit, and the apology qualifier described in the issue. Done means qualifying low-effort help requests receive the new response without disrupting longer questions.

Written by the indexing model from the issue text.