[Iris] Fix stale detail state and nested child jobs in dashboard

[ahmeda14960]

Fix route-reuse stale state in task, worker, log, and job detail views, and show nested child jobs from job detail without leaving the page. Share the job tree logic with JobsTab and keep direct children as the default detail-page view with an all-descendants toggle.

Fixes #4107

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 74bd8a8c52

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Preserve queued duration for child jobs

The child-jobs table now renders duration via jobDuration(node.job), which returns '-' until startedAt is set. This drops the previous queued-time fallback (queued <relative time> from submittedAt) for pending child jobs, so users lose visibility into how long queued children have been waiting. That is a behavioral regression in the detail page’s status signal for unscheduled child jobs.

Useful? React with 👍 / 👎.

[rjpower]

@claude Looks like something in the change is breaking the cloud smoke tests expected playwright rules. pls fix. Also please remove the debugging log doc from this branch

[claude]

Claude finished @rjpower's task in 5m 0s —— View job

---

Fixing Playwright smoke tests and removing debug doc

• Gather context: CI failure is test_dashboard_jobs_tab — sleep-30 job finishes (SUCCEEDED) before wait_for_state(RUNNING) can observe it. This is a timing flake in cloud mode where scheduling takes longer.
• Remove docs/debug-log-dashboard-route-reuse.md
• Fix test_dashboard_jobs_tab: increase sleep duration from 30s → 300s so the job stays RUNNING long enough for the test to observe it (the job is killed at the end of the test anyway)
• Run pre-commit — all checks pass
• Commit and push

Summary

Root cause: In cloud mode, wait(quick) + wait(failed) consumed enough wall time (due to GCE scheduling latency) that the 30-second sleep job completed before wait_for_state(RUNNING) was called, causing a TimeoutError (job was already SUCCEEDED).

Changes:

1. Deleted docs/debug-log-dashboard-route-reuse.md (debugging artifact)
2. Increased sleep duration in test_dashboard_jobs_tab from 30s to 300s — the job is killed at the end of the test, so the longer sleep has no impact on test duration

Testing: Pre-commit passes. The actual cloud smoke test will validate on the next CI run.
·