ci: retry MSan apt updates after transient repository errors

[DeusData]

APT can report a transient repository download failure while returning success. That prevented the MSan image's existing retry helper from retrying apt-get update, leaving the following toolchain installation without the required package index.

Pass --error-on=any to all three update commands so those failures enter the existing bounded retry loop. The option is documented in the Ubuntu Noble apt-get manual.

Validation used the Dockerfile's pinned Ubuntu image and its actual retry script/update arguments against an inaccessible loopback repository, with container networking disabled:

• Before the fix: all three updates returned success after one attempt; the regression check failed.
• With the fix: all three exhausted five attempts and failed as required; the regression check passed.
• Removing the flags reproduced the failure; restoring them passed again.

The complete MSan image build and remote CI remain to be verified.

[DeusData]

Checkpoint and handover (2026-10-03 UTC)

Published head: 5692dbe29982a344ee2d712957f75ece23a6557e. The remote head was verified.

The APT correction has historical fixed/reverted/restored proof. Hosted run 37051012678 / job 111160434967 passed 8,155 native tests, then artifact upload failed with ENOTFOUND. One selective rerun was rejected as unavailable. Preserve the infrastructure attribution and check run state before proposing another action; the remote gate remains unresolved.

Hosted snapshot at 2026-10-03 21:54:00 UTC: 1 failure, 1 queued, 3 skipped, 28 success. Failed checks: test / test-unix (macos-15-intel, cc, c++, 1/1). Confirm the required checks on this exact head before treating it as ready.

No additional local build, test, lint, sanitizer, benchmark or CI runs were performed at this checkpoint, as requested. Earlier executed evidence remains historical; prepared tests and the newer source-reviewed changes must still be validated by the hosted gate.

The campaign is paused at the maintainer’s request. Local monitoring has stopped; hosted jobs remain running. No merge was performed. Thanks for reviewing this change.