Diagnostic message for retried tests

[Trott]

This is dependent on #3941 being landed first.

Tests on SmartOS are sometimes retried due to a SmartOS issue on CI.
When this happens, a TAP diagnostic message is written.

R=@indutny

[Trott]

CI: https://ci.nodejs.org/job/node-test-pull-request/804/

[indutny]

Fantastic! LGTM

[Trott]

TL;DR: Ran a few tests to make sure it works as advertised on SmartOS with ECONNREFUSED errors and it looks good.

I doctored up a test that would always fail with ECONNREFUSED and ran it through CI. Here's what it looks like:

not ok 209 test-econnrefused.js
# ECONNREFUSED received, test retried
# events.js:142
#       throw er; // Unhandled 'error' event
#       ^
# 
# Error: connect ECONNREFUSED 127.0.0.1:12346
#     at Object.exports._errnoException (util.js:873:11)
#     at exports._exceptionWithHostPort (util.js:896:20)
#     at TCPConnectWrap.afterConnect [as oncomplete] (net.js:1066:14)
  ---
  duration_ms: 0.549

See https://ci.nodejs.org/job/node-test-commit-smartos/404/nodes=smartos14-64/console for the full console log.

Then I made it so it would fail with ECONNREFUSED the first time it ran but succeed the second time it ran. Here's what that looks like:

ok 209 test-econnrefused.js
# ECONNREFUSED received, test retried
  ---
  duration_ms: 0.548

For the full console log on that one: https://ci.nodejs.org/job/node-test-commit-smartos/406/nodes=smartos14-64/console

[indutny]

Still LGTM ;)

[Trott]

One last post-rebase CI: https://ci.nodejs.org/job/node-test-commit/1233/

[Trott]

CI failures on arm-fanned and smartos but they both look Jenkins-related and not actual test failures. /cc @nodejs/build

Failures on windows-fanned look unrelated too. I'll confirm previous failures for those two tests and get them added to the flaky test list. :-/

[Trott]

Landed in 41519fd