fix: Bugbot findings from promotion PR #21

[shujaatTracebloc]

Resolves the two Cursor Bugbot findings surfaced on the start-training promotion PR (#21, develop → main).

Fixes

• Clone failure misreported as success — the model-zoo clone cell discarded stderr (2>/dev/null) and ran echo "model-zoo already cloned" on any non-zero exit, so network errors, permission problems, or a partial directory looked like success and later cells failed without a clear cause. Now it checks whether ../model-zoo actually exists; if not, it runs git clone and lets the real error surface. (chore(release): promote develop → main (week of 2026-06-29) #21 (comment))
• WIP check skips reopened PRs — wip-limit-check.yml listened only for opened/ready_for_review, so a reopened PR re-entered active work without a WIP-limit check, unlike add-to-kanban.yml/set-pr-status.yml. Added reopened. (chore(release): promote develop → main (week of 2026-06-29) #21 (comment))

Lands on develop; the promotion PR head picks it up on the next sync and Bugbot re-reviews.

🤖 Generated with Claude Code

---

Note

Low Risk
Small CI trigger alignment and a notebook shell one-liner; no auth, data, or production runtime paths.

Overview
Addresses two Bugbot findings from the develop→main promotion PR.

WIP limit on reopened PRs: wip-limit-check.yml now runs on reopened as well as opened and ready_for_review, matching add-to-kanban.yml and set-pr-status.yml so reopened PRs get a WIP check again.

Training notebook model-zoo clone: The Option A cell no longer uses git clone … 2>/dev/null with a fallback message that treated any failure like “already cloned.” It checks whether ../model-zoo exists first; only if missing does it run git clone, so real clone errors surface instead of misleading success.

^{Reviewed by Cursor Bugbot for commit 1421051. Bugbot is set up for automated code reviews on this repo. Configure here.}