chore(buzz-acp): repin the fork to the DCO-signed commit of block/buzz#6088

[jhgaylor]

Follows the fork pin through the DCO fix on block/buzz#6088. The PR's original commit was amended with Signed-off-by (83eda69a → af535a64, identical tree), and a second commit 9390060 adds the reviewer-requested README docs for BUZZ_ACP_STATE_DIR / --no-resume-sessions. The old sha is unreachable from any branch on the fork, so the pin moves before GitHub can GC it. #776 still tracks the eventual repin to upstream.

🤖 Generated with Claude Code

[jhgaylor]

Closing: superseded — the DCO-signed commit this repins to is already in main's lineage.

When this PR was opened the pin was 83eda69 (unsigned) and this moved it to 9390060. Since then #787, #788 and #792 have repinned the fork three more times, and main now sits at 5bbad010 (fountain.4). Checking that against the signed commit:

• af535a64...5bbad010 → status=ahead, ahead=5, behind=0 — the DCO-signed af535a64 is an ancestor of the current pin.
• 83eda69...5bbad010 → status=diverged, behind=1 — the unsigned commit is not.
• Every commit between them carries a Signed-off-by, including 9390060 itself.

So the fork history was rebuilt onto the signed commit and main followed it there by another route. Merging this now would move the pin backwards from 5bbad010 to 9390060, dropping five commits (!rotate fresh-session handling, owner control commands with rendered mentions, and the kind:10100 relay-agent directory entry that fountain.4 exists for) — and, because buzz-acp-publish.yml runs gh release upload --clobber against the already-published buzz-acp-v0.5.14-fountain.4, it would overwrite that release's assets with the older build.

Nothing to salvage here; #776 still tracks repinning to upstream once block/buzz#6088 lands.