fix(search-sudoku-gt): BETA->PROD, convention #1946 informative prints (14 notebooks)

[jsboige]

Promote 14 notebooks from BETA/ALPHA to PRODUCTION by adding informative print statements to every silent code cell (convention #1946). Papermill-verified execution (0 errors, 0 missing outputs). Catalog regenerated.

[github-actions]

Notebook PR Validation: PASS

• Notebooks checked: 14
• Code cells validated: 217
• Result: All passed

Checks: H.1 (no errors), H.3 (execution_count), C.1 (no banned patterns)
Non-Python kernels (.NET/Lean): C.1 + errors only (execution_count advisory)
QuantConnect notebooks: C.1 + errors only (require QC Cloud for execution)

[jsboige]

[NanoClaw]

Verdict: APPROVE with 1 cosmetic fix recommended

Systematically reviewed all 24 changed files (20 listed + 4 unlisted in GitHub diff API).

---

1. Security scan -- PASS

• Zero secrets, API keys, tokens, or credentials in the diff.
• Zero emails or personal info leaks.
• Pre-existing localhost:8080 in Search-6 (kernel base_uri, not added by this PR).

2. Notebook output quality -- PASS (14/14 notebooks sampled)

Every notebook at HEAD has execution_count on all code cells (zero null). All cells that should have outputs do have outputs. No fake/N/A/QuantBook outputs detected.

Notebook	Code cells	With outputs (base->head)	Convention prints
GT-3 Topology2x2	15	13->14	1 (Imports OK)
Search-2 Uninformed	29	21->29	8
Search-6 Adversarial	14	11->14	2
Sudoku-1 Backtracking	11	10->11	2
Sudoku-8 HumanStrategies	19	13->19	6
Sudoku-18 Comparison	21	18->21	5
GT-2,4,5,6,9,10,11,14	11-16 each	+1 each	1 each

3. GameTheory-3 expansion (+716/-73) -- LEGITIMATE

The large diff stat is misleading. Actual changes: +1 markdown cell (conclusion), +1 print in cell 0, JSON key reordering in outputs from re-execution. Total JSON size actually shrank (166KB->165KB). No new code logic added.

4. Catalog consistency -- PASS (with note)

JSON: PRODUCTION 357->371 (+14), BETA 107->93 (-14). MD: PRODUCTION 160->371 (+211), BETA 297->93 (-204).

The MD catalog was severely out of sync with JSON at the base commit (PRODUCTION: 160 vs 357). This PR corrects that discrepancy. The 14 promotions account for the JSON delta; the remaining MD delta is a long-overdue reconciliation.

README badges updated consistently:

• GameTheory: BETA=9 -> PRODUCTION=23 (correct: 14 original + 9 promoted)
• Search: PRODUCTION=39->41, BETA=5->3 (2 promotions)
• Sudoku: PRODUCTION=22->25, BETA=10->7 (3 promotions)

5. BETA->PROD promotions -- WARRANTED

All 14 promoted notebooks have cells_without_outputs: 0 and recent last_validation dates (May 2026). The convention prints give every code cell an observable output, satisfying the all-have-outputs maturity gate from PR #1946.

6. Convention prints -- GOOD

Follow the PR #1946 convention consistently: print("Imports OK : ...") for imports, print("Classe X definie : ...") for definitions, print("Exercice a completer : ...") for stubs. All informative, no noise.

7. Stealth files (not in GitHub file listing) -- NOTE

4 SymbolicAI notebooks also changed in the commit but not listed in the GitHub PR diff API:

• Linq2Z3.ipynb: Removed leaked internal IPs (fe80::, 172.x, 192.168.x) from dotnet-interactive probeAddresses. Removed fake "Installed Packages" output. Clean re-execution.
• Planners-5-Heuristics.ipynb: Removed placeholder print, added real execution outputs. Set-dict ordering changes (benign).
• Planners-10-LLM-Planning.ipynb: Removed placeholder print. Convention-consistent.
• SC-14-Formal-Verification.ipynb: Removed placeholder print. Convention-consistent.

These follow the same #1946 convention. Consider listing them in the PR title/description for transparency.

8. One issue to fix

GameTheory-11-BayesianGames.ipynb: Cells 5 and 6 have user-visible stderr outputs containing the local path:

C:\Users\jsboi\AppData\Local\Temp\claude\ipykernel_46512\2051189024.py:37: RuntimeWarning

This is a student-facing notebook going to PRODUCTION. The RuntimeWarning itself is fine (scipy convergence), but the local temp path leaks the author's machine. Recommend sanitizing these stderr outputs (strip or replace the path prefix) before merge. This was also present at the base (as /tmp/ipykernel_456/...) so it's a pre-existing issue, but this PR is the right time to fix it since the outputs are being regenerated.

Also note: papermill metadata in all 14 notebooks contains output_path: C:/Users/jsboi/AppData/Local/Temp/.... This is not student-visible but is a personal path in the repo. Pre-existing pattern.

[jsboige]

Fermeture deconfliction (collision cross-instance GameTheory, 2e occurrence cette nuit). Aucune perte de travail — le contenu de cette PR est integralement couvert ailleurs :

• GameTheory (GT-2/3/4/5/6/9/10/11/14) -> fix(GameTheory): add convention prints #1946 to 10 notebooks (9 BETA->PRODUCTION) #1956 (po-2025:CoursIA-2, lane GameTheory/.NET-CPU, superset incl. GT-4c)
• Search-2/6 -> chore(search): re-execute 10 Search notebooks with Papermill #1944 (po-2024:CoursIA-2, deja rebase+catalog)
• Sudoku-1/8/18 -> chore(gametheory): re-execute 9 notebooks with fresh papermill outputs #1939 (deja merge sur main)

Rappel partition (anti-collision, 1 editeur/serie) : GameTheory + Sudoku + Search = lane CoursIA-2 UNIQUEMENT. Avant d'editer un notebook de ces series : gh pr list + grep du path pour verifier qu'aucune PR ouverte (toutes instances) ne le touche deja. Merci !