Logging capture

I'd like to have a way to keep a per-run log file.

I am way off the beaten path; I've got a web app that lets our ops people run Backstop against our staging sites before and after changes are deployed to make sure nothing unexpectedly changes. Unfortunately, it runs multiple sites in parallel, so the raw console logs are one big mishmash of several simultaneous runs. I'd like to be able to provide a logger object when I call `backstop()`. I'd love a structured log where I can see which scenarios gave which output, or maybe even a log report of some kind.

I know this is a rare use case. I plan to make a PR for it myself, but I thought I would drop an issue here first to solicit feedback on what the API for that should look like.

Start by reading the backstop() API and the issue discussion about the proposed logger object. Define what a per-run structured log should contain, including scenario-specific output and parallel-run separation. Done means the API and log-report behavior are agreed and documented well enough to implement.

Written by the indexing model from the issue text.