Background

Having done some testing, I'm pretty convinced it's unlikely users are easily hitting the 5000 request per-hour quota. I think what's instead happening is a secondary quota that occurs, either:

• due to too many concurrent requests (which can be controlled with --concurrency), or.
• due to too many requests in a short period of time (perhaps due to fast home Internet?), or.
• perhaps some ISPs are themselves triggering a 403?

In pursuit of unblocking people

1. this PR defaults the delay between requests to 500ms, and introduces a jitter, such that even when running requests in parallel there should be some separation between them.
2. feat: introduce retry and delay configuration #563 introduces retry logic (enabled via --retry) which when turned on will retry spurious 403s several times. It also documents --concurrency which I recommend adjusting to see if 403s stop cropping up.
3. feat: detailed debugging now configured with NODE_DEBUG=repo #567 adds logging of GitHub ratelimit headers, enabled via NODE_DEBUG=repo repo foo, I suggest users bumping into issues set this environment variable to confirm the hypothesis about the source of 403s.

Refs: #556