[ItIsUday]

Why

The CLI resolves JUnit <testsuite timestamp> timezones itself with a blanket tzlocal(). That's correct for offset-aware and genuinely-local frameworks, but wrong for "looks-naive-but-UTC" runners (jest, cypress), corrupting test_cases.created_at on non-UTC agents and breaking cross-machine elapsed-time aggregation (AIENG-379, Option 2). Mirroring the framework-timezone policy in every client is also unmaintainable.

What

Move timezone resolution to the server (single source of truth). The CLI now:

• sends the raw <testsuite timestamp> string as suiteTimestamp (unresolved);
• sends the agent's UTC offset once per record tests request (agentUtcOffset) — only the client knows it, and the server needs it to resolve naive local timestamps.

_timestamp()/tzlocal() and the framework-TZ imports are removed — the CLI holds zero timezone knowledge. Requires the paired mothership change that accepts these fields and resolves them.

Tests

Full suite green except one pre-existing, unrelated subset failure (test_maven.test_scan_test_compile_lst, fails on clean main). Updated golden payloads to the raw-passthrough format; comparator now excludes the machine-dependent agentUtcOffset.

[ono-max]

@ItIsUday

A lot of customers are still on CLI v1, so can we apply this change to v1 as well? Here is the process to apply the change to both versions: https://cloudbees.atlassian.net/wiki/x/u4FcUAE

[ItIsUday]

Superseded: per the v1→main process this fix must land on v1 first (#1413) and reach main via forward-port (git merge v1), not as a direct main PR. Closing; the main side will be the forward-port of #1413.