fix(rules): restore pk column to Rules CSV (follow-up to #228)

[jsboige]

Problem

PR #228 merged with DatasetUpdaterRootConfig.cs referencing PrimaryField = "pk" for the Rules PT translation config, but the accompanying CSV change (pk column addition with Rules_01..Rules_18 prefixes) was lost during an earlier rebase of that PR branch.

Master's Cards/Rules/Argumentum Rules - Cards.csv currently has headers:
```
Text,Text_en,Text_ru,Text_pt,print_and_play
```

No `pk` column. The config references it anyway:

```csharp
// DatasetUpdaterRootConfig.cs:675
PrimaryField = "pk",
```

Result: if someone sets Enabled = true on the Rules PT config and runs the pipeline, DatasetUpdater will fail to identify rows (no primary key column to match).

Fix

Restore the exact pk column that was in the original PR #228 commit (`a99f228a`), matching what Rule.cs expects via its Pk property + ClassMap.

Header becomes:
```
pk,Text,Text_en,Text_ru,Text_pt,print_and_play
```

Rows get prefixed Rules_01..Rules_18. 18 data rows — matches the game's rule-card structure (École des Menteurs, Bingo, Parlote Coinchée variants).

Scope

• One file changed: Cards/Rules/Argumentum Rules - Cards.csv
• No code changes — Rule.cs already has .Optional() on the Pk mapping (robust to reads)
• Config still Enabled = false — this is purely a latent-bug fix with zero runtime behavior change for anyone who hasn't enabled the config

Diff

19 insertions + 19 deletions (just the row-level boundaries where pk is prepended):

```
-Text,Text_en,Text_ru,Text_pt,print_and_play
+pk,Text,Text_en,Text_ru,Text_pt,print_and_play
-"# Argumentum
+Rules_01,"# Argumentum
...
```

Validation

• ✅ Build: 0 errors, 17 warnings (all pre-existing)
• ✅ Tests: 88 pass, 1 skip (Freeplane GUI — interactive-only), 0 fail

Test plan

• Verify Rules CSV reads cleanly via existing RuleClassMap (Pk mapping with .Optional() already in place)
• Manual: temporarily Enabled = true on Rules PT config with TakeChunkNb = 1 and confirm DatasetUpdater identifies rows by pk (full end-to-end run deferred to translation campaign with OpenAI key)

Related

• PR feat(dataset-updater): add Rules PT translation config #228 — added the DatasetUpdater config + Rule.cs Pk property, but lost the CSV change in rebase
• PR feat(dataset-updater): Scenarii FR→EN/RU/PT translation configs (#219) #232 (open) — Scenarii translation configs, unaffected by this fix

🤖 Generated with Claude Code

[jsboige]

Review — self-review before merge

Contexte

Fix chirurgical pour un bug latent découvert après merge de #228. Je l'ai documenté en détail dans la PR body. Rappel rapide :

• DatasetUpdaterRootConfig.cs:675 sur master a PrimaryField = "pk" pour la config Rules PT
• Cards/Rules/Argumentum Rules - Cards.csv sur master n'a PAS de colonne pk (perdue lors d'un rebase de feat(dataset-updater): add Rules PT translation config #228)
• → Si quelqu'un Enabled = true sur la config, DatasetUpdater ne trouverait pas de clé primaire → crash

Audit du fix

Item	Vérification
Colonne pk ajoutée	✅ pk,Text,Text_en,Text_ru,Text_pt,print_and_play
Préfixes Rules_NN	✅ Rules_01..Rules_18 (18 lignes = 18 cartes règles)
Correspond au commit original #228	✅ Restauration exacte du contenu a99f228
Rule.cs côté code	✅ Pk mapping déjà présent sur master avec .Optional()
Impact runtime tant que Enabled=false	✅ Zéro (config désactivée par défaut)

Validation

• ✅ Build: 0 erreurs
• ✅ Tests: 88 pass / 1 skip (Freeplane GUI) / 0 fail
• ✅ GitGuardian Security Checks: SUCCESS
• ✅ Mergeable: CLEAN
• ✅ Diff minimal: 1 fichier, +19/-19 (prépendage de colonne uniquement)

Conclusion

Fix safe, self-contained, resolves a latent bug. Prêt à merger.

[clusterManager-Myia]

Post-hoc review — PR #233 fix(rules): restore pk column to Rules CSV (follow-up to #228)

Verdict: ✅ LGTM — necessary follow-up

This PR restores the pk column to the Rules CSV that was referenced by PR #228's DatasetUpdater config but missing from the actual CSV file.

Observations:

• PK values follow the Rules_01, Rules_02, ... pattern — consistent and clear.
• All 18+ rows get proper PK assignment.
• No content changes to any text columns — purely structural.
• This is the kind of follow-up that highlights the value of post-merge review: the CSV config was deployed before the CSV schema was updated.

No regressions. Correct fix.