PR reviews: How to handle related but irrelevant ideas / brainfarts

[EwoutH]

When reviewing a PR, you often think about the larger implications of that PR, or you noticed things or thought about things that might be useful but irrelevant to this PR. Those ideas can be very useful and important on the long term, but they can hold up PRs by being discussions not relevant for an atomic PR.

It's a huge opportunity: On a code change you often also review related code or higher level content, which might have not been thought of in a while. We need a good system to create spinoff issues/discussions from that, to keep the good stuff but don't bloat our PRs.

One complication is that you might not want to create a new issue/discussion for every little initial brain fart, because at that point you don't know if it's a good idea or not. But you also don't want to have one thousand spinoff discussions in a PR. What's a good balance or process here?