fix(ci): timeout remaining llms-full and r2-pages fetches

[SebTardif]

What Problem This Solves

llms-full.yml still does live fetch without AbortSignal. r2-pages.yml token verify uses curl -fsS with no --max-time, and the credential mint fetch has no abort. Sibling live-smoke and cutover scripts already bound those calls.

Evidence

Terminal output from the workflow contract checks:

$ node --test scripts/docs-site/llms-full-workflow.test.mjs scripts/docs-site/r2-pages-workflow.test.mjs
✔ every inline llms-full live fetch uses AbortSignal.timeout
✔ r2-pages credential mint curl uses --max-time
✔ r2-pages credential mint fetch uses AbortSignal.timeout
ℹ tests 3
ℹ pass 3

Real behavior proof

• Behavior or issue addressed: Remaining docs publish fetches could hang until the job timeout.
• Real environment tested: Windows 11, Node, docs workflow files on current main.
• Exact steps or command run after this patch:

node --test scripts/docs-site/llms-full-workflow.test.mjs scripts/docs-site/r2-pages-workflow.test.mjs

• Evidence after fix: terminal output from the patched tree:

✔ every inline llms-full live fetch uses AbortSignal.timeout
✔ r2-pages credential mint curl uses --max-time
✔ r2-pages credential mint fetch uses AbortSignal.timeout

• Observed result after fix: Both llms-full live fetches use AbortSignal.timeout(30_000). Token verify curls use --max-time 30. The temp-access-credentials fetch uses the same abort signal.
• What was not tested: A live GitHub Actions run against docs.openclaw.ai.

Summary

Bound the remaining llms-full and r2-pages network calls.