fix(cli): use --timeout-seconds for cron payload timeout

[funsaized]

Summary

Fix CLI fallback timeout flag when creating cron jobs from ensure-crons.

Problem

Antfarm maps job.payload.timeoutSeconds to OpenClaw CLI --timeout, but OpenClaw interprets that flag as gateway/client timeout in milliseconds.

This means a value like 120 (intended seconds) becomes 120ms, causing failures such as:

• gateway timeout after 120ms
• gateway client stopped

Fix

Use --timeout-seconds instead of --timeout when passing job.payload.timeoutSeconds.

Validation

• Reproduced failure with --timeout 120
• Verified success with --timeout-seconds 120
• Verified antfarm workflow ensure-crons wedding-speech-research recreates all expected jobs

Closes #307

[vercel]

Someone is attempting to deploy a commit to the Ryan Team on Vercel.

A member of the Team first needs to authorize it.

[funsaized]

Update after validating OpenClaw CLI semantics:

• --timeout = gateway/client request timeout (milliseconds)
• --timeout-seconds = agent job payload timeout (payload.timeoutSeconds)

This PR intentionally maps Antfarm job.payload.timeoutSeconds to --timeout-seconds so CLI fallback matches HTTP behavior and correctly sets job payload timeout instead of transport timeout.