another test flake after failure to sync NTP in helios-deploy

[davepacheco]

This looks just like #4307. Here's the test failure:

https://github.com/oxidecomputer/omicron/pull/5345/checks?check_run_id=23226775903
https://buildomat.eng.oxide.computer/wg/0/details/01HT43K6DFF7KS1M2PN7KQFTDG/fW9lP6upEQI9pj7K7lEZocNM6gn0czizN5qB8WHkeoWga25U/01HT43Q01QEF77V3S3V6MNGSVY

Similarly, sled agent reports "Time is not yet synchronized" from 05:53:22.154Z til 05:58:17.467Z. We gave up on the test at 05:58:31.781Z.

The PR change looks pretty unrelated to any of this.

[jgallagher]

https://github.com/oxidecomputer/omicron/pull/5373/checks?check_run_id=23319866090
https://buildomat.eng.oxide.computer/wg/0/details/01HTDV7XRQHZRRKCXA99M07MT5/bs5ryrEmko0wbeGKKquaWLq3t8VwgqGE62GcAERf1OkJZ8Rx/01HTDV8D5TNY83QWBZ0CDZXW7W

looks like the same failure, on a PR that definitely did not change anything pre-Nexus.

[ubedan]

This has been reported as a test flake, however I'm seeing it consistently.

[bnaecker]

Saw this again on #6925, here is a link to the sled-agent log showing time failed to sync.

[davepacheco]

Saw this again on #7122: failure, sled agent log.

[smklein]

Saw this again on https://github.com/oxidecomputer/omicron/pull/7008/checks?check_run_id=34558916115 , sled agent logs here: https://buildomat.eng.oxide.computer/wg/0/artefact/01JFB5R7Z4M7ED39D6G37RYVZ8/8VtttZi7cnjzazucMG9t17pR9wQrVj8WxS6y0Dd8i8sCVS9j/01JFB5SGVRMBN7RM8HDGB9KN0Y/01JFBAAXG7WM4M2Q1NXNM703EV/oxide-sled-agent:default.log?format=x-bunyan