Way to surface non-critical errors in the report

[ebidel]

From #1492 (review)

For individual gatherers/audit errors, we surface debugString.

Right now, fatal errors stop the extension from running and we show a "report error" button. However, for non-fatal errors that happen outside of the gatherer -> audit -> report pipeline, we don't have a way to surface those errors to extension users.

Without analytics, we don't know about non-fatal errors. The HTML report could have a section where users see overall LH non-fatal errors and file bugs against us.

[brendankenny]

It seems this could be broken down into

• a general system for logging or capturing errors in Lighthouse
• having those errors available for output in some way: surfacing in the report, error reporting if that's enabled, etc

[brendankenny]

after #1591, we probably want to add the errors in the gatherer -> audit -> report pipeline to this as well. Some things might not make sense, but e.g. if we couldn't find FCP and didn't calculate a bunch of perf metrics it should probably be called out at the top as well (this was the spirit of @addyosmani's original #812)

We may want a better icon to mark them in the report as well :)

[brendankenny]

one more thought for the CLI: as @patrickhulce notes in #1605 (comment), if something gets logged at the beginning of the LH run, it probably won't get noticed. If we figure out a nice way to pipe errors to the report (or whatever), we could also drop them in the log at the end of the CLI run so they're very visible there too.

[brendankenny]

Starting list of things to warn on:

• Run on localhost, so is-on-https warning (amongst others?) might not be representative of when site moves to production/real server - (formerly Skip "production" audits when in development #3228)
• redirected from HTTP to HTTPS (should only happen in the CLI), so is-on-https passes (after is-on-https and insecure redirects #3417) but risk of downgrade attack and performance impact of redirect (details in is-on-https and insecure redirects #3417)

[patrickhulce]

addressed by #3692