docs(epics): refresh roadmap with pilot v1 / MENA-10 / launch release phases

[abdout]

Why

The existing 28-epic / 1115-pt production roadmap (.bmad/epics/epic-prod-overview.md, ~16 sprints / 32 weeks) treats every feature equally and predates the live pilot. Today reality is more urgent:

• Pilot Seat 1, King Fahad Schools (Sudan, K-12), is on the free tier since Apr 2026; conversion conversation Jun 2026.
• MENA-10 target: 10 free pilots signed by Aug 31, 2026; ≥1 paid conversion by Oct 31, 2026.
• The app is a UI shell on top of stubs (2/92 endpoints, broken mutation queue, fake SubmitGradeUseCase, 22/27 modules untested).

Without a release-phase overlay, "what ships next" requires re-deriving priorities from 28 epic files each time. The refresh fixes that.

What

Add a release-phase axis to the existing roadmap (epic IDs preserved):

Phase	Sprints	Weeks	Pts	Target
1 — Pilot v1 (King Fahad)	P1–P3	7	~296	2026-07-12
2 — MENA-10	P4–P8	10	~411	2026-09-15
3 — Public Launch (v1.0.0)	P9–P16	16	~408	2026-12-20

Build phases (A Foundation → F Production) describe dependency order. Release phases (1/2/3) describe ship order. Orthogonal.

Deliverables

• .bmad/epics/README.md — two-axis model + index
• .bmad/epics/phase-1-pilot-v1.md — 71 stories, sprint cadence, DoD, deferrals
• .bmad/epics/phase-2-mena10.md — teacher write, real-time, Stripe, OAuth, foundation cleanup
• .bmad/epics/phase-3-launch.md — LMS, library, wallet, accessibility, 80% coverage, Play Store production
• .bmad/epics/epic-prod-overview.md — Release Phases section added
• .bmad/bmm-workflow-status.yaml — release_phases, phase_1_story_ids, release_phase_concentration
• docs/prd.md — Milestones rewritten around the three release phases
• Legacy V1 (epic-01..17) + V2 (epic-v2-01) + docs/epics/* archived to .bmad/epics/archive/

Phase 1 scope (the load-bearing decision)

71 stories from across 15 epics. Pilot ships:

• Auth incl. biometric + school selector
• Read-only views: grades, attendance, timetable, fees, announcements
• FCM data-channel push + deep links
• Full admission flow (form, list, status, tours, events) — sales-demo critical
• Arabic-complete with native-speaker translation audit
• Signed APK on Play Internal Testing track + Crashlytics + Performance + Analytics

Explicitly deferred (per-epic in phase-1-pilot-v1.md): teacher write, Socket.IO messaging, Stripe payment, Google/Facebook OAuth, LMS, library, ID card / wallet, quiz, public Play Store, full RBAC ability layer.

Acceptance

• Phase manifests + README readable + cross-linked
• epic-prod-overview.md has Release Phases section
• bmm-workflow-status.yaml parses
• Legacy V1 + V2 + docs/epics/* moved to .bmad/epics/archive/
• Captain (@abdout) + Samia review and confirm 7-week timeline for King Fahad conversion is realistic

Risks

1. Server endpoint slippage (hogwarts repo) — half of E08 is web-team work; track per-endpoint owners
2. Native-speaker Arabic audit — engage Crowdin/Lokalise vendor in P1 week 1
3. Conversion conversation may happen before P3 ships — demo from P2-end build (week 5)
4. Velocity is hot (88/110/98 vs target 50–80) — bail-out: drop E04.S01 audit if needed
5. Play Console deletion-policy review can take 2–4 weeks — submit listing draft P1 week 1

Source plan: ~/.claude/plans/take-web-app-as-precious-turtle.md.