chore(pii): drop beta-3 tester surname; align beta-2 numbers to 2026-07-03 snapshot

Replace the residual surname with the betaD pseudonym in the cv-structure
skill and the two 2026-07-01 postmortems (5 occurrences), and align the
public numbers (README + RESULTS case #4) to the live 2026-07-03 snapshot:
658 found / 520 scored (avg ~71) / 307 strong (>=70) / 169 excellent (>=80).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

Author: leopu00

README.md

@@ -40,11 +40,11 @@ I originally built JHT for my own job hunt. It worked. So I rebuilt it as open s
 
 > 📊 **From the original private build** — ~200 offers analyzed · ~20 tailored applications · **5 interview invites within a few weeks**. Full background in [`docs/about/STORY.md`](docs/about/STORY.md).
 
-> 📈 **From the public stack (June 2026)** — a Codex-powered team ran **one month unattended**: **649 positions found · 513 scored**, closing its weekly budget at **99–100% for four straight weeks** with zero human interventions. Data in [`docs/about/RESULTS.md`](docs/about/RESULTS.md), live on [jobhunterteam.ai/case-studies](https://jobhunterteam.ai/case-studies).
+> 📈 **From the public stack (June 2026)** — a Codex-powered team ran **one month unattended**: **658 positions found · 520 scored · 307 strong matches (score ≥70)**, closing its weekly budget at **99–100% for four straight weeks** with zero human interventions. Data in [`docs/about/RESULTS.md`](docs/about/RESULTS.md), live on [jobhunterteam.ai/case-studies](https://jobhunterteam.ai/case-studies).
 
 ## 🎬 Demo
 
-The best demo is real data: [jobhunterteam.ai/case-studies](https://jobhunterteam.ai/case-studies) renders the live dashboards of the beta runs — including the month-long autonomous Codex run (649 positions found, 513 scored, weekly budget self-managed at 99–100%).
+The best demo is real data: [jobhunterteam.ai/case-studies](https://jobhunterteam.ai/case-studies) renders the live dashboards of the beta runs — including the month-long autonomous Codex run (658 positions found, 520 scored, weekly budget self-managed at 99–100%).
 
 > **Where the numbers come from** — every dashboard renders an anonymized snapshot of the team's event log, committed to this repo ([`web/data/case-studies/`](web/data/case-studies/)). Same data, same charts, reproducible from the JSON. Methodology and per-run breakdowns in [`docs/about/RESULTS.md`](docs/about/RESULTS.md).
 
@@ -141,7 +141,7 @@ Three subscriptions cover the ~400M tokens/month requirement:
 | | Provider | Plan | Cost/mo | Status |
 |---|---|---|---|---|
 | 🟠 | **Claude** | Max x20 | ~€200 | ✅ Production-ready, best precision |
-| 🔵 | **Codex** | Plus / Pro | ~€100 | ✅ Proven — 1-month autonomous run (649 positions, weekly budget self-managed at 99–100%) |
+| 🔵 | **Codex** | Plus / Pro | ~€100 | ✅ Proven — 1-month autonomous run (658 positions, weekly budget self-managed at 99–100%) |
 | 🌙 | **Kimi** | Pro | ~€40 | 🧪 Beta — mass-market tier (75h + 10-day runs; two multi-week teams in observation) |
 
 ---

docs/about/RESULTS.md

@@ -131,11 +131,11 @@ Raw data: SQLite snapshot, Sentinel logs (3052 ticks), Kimi session logs (16,700
 |---|---|
 | 👤 User profile | Early-career finance (credit risk → front office) |
 | 🌍 Target geography | European financial hubs |
-| 📅 Period | 2026-06-03 → ongoing (**28 days** covered at the 2026-07-01 snapshot) |
+| 📅 Period | 2026-06-03 → ongoing (**30 days** covered at the 2026-07-03 snapshot) |
 | 💳 Subscription | 🔵 **Codex Pro ~€100/mo** |
-| 🎯 Job offers found by the pipeline | **649** |
-| ✅ Scored | **513** (avg ~70/100 at the mid-run snapshot, range 27–94) |
-| 💪 Strong matches (score ≥70 / ≥80) | **303 / 166** |
+| 🎯 Job offers found by the pipeline | **658** |
+| ✅ Scored | **520** (avg ~71/100, range 27–94) |
+| 💪 Strong matches (score ≥70 / ≥80) | **307 / 169** |
 | 📄 Applications | on-demand only *(writer-on-demand: the user orders CVs when they want — this run measures the search & scoring engine)* |
 | 💰 Weekly budget utilization | **≈100% every week, zero overshoot** (4.1 weekly budgets consumed in 4 calendar weeks) |
 | 🧠 Human interventions | **0** — observed read-only for the whole month |