docs(domain): business legend source of truth (plan step A1)

[brownjuly2003-code]

What

New docs/domain.md — the single source of truth for the demo business legend:

• The company: mid-size Russian own-brand (private-label) importer of small kitchen appliances — RU design/QC, contract manufacturing in China, ~2,000 orders/day / ~3–5 B ₽/yr, market-grounded against the public own-brand importer archetype.
• Footprint: 5 branches / 3 jurisdictions, incl. dxb as a JAFZA free-zone re-export hub (goods outside UAE customs territory, 0% qualifying free-zone income) and ala as the EAEU contour.
• Channels: B2B wholesale (~150–200 orders/day, 30–80k ₽, retro-bonuses, Bitrix24 CRM) + FBS marketplaces (~1,800/day, 1.5–3k ₽) + D2C site — deliberately bimodal economics.
• Seasonality & compliance: gift-driven peaks, 40–60-day China containers, Chestny ZNAK marking (mandatory for importers since 2026-05-01) grounding hub_marking_code.
• Reinterpretation dictionary (§5): every existing serving entity / metric / DV2 hub / satellite / record_source prefix read in domain terms. Ground rule: no code renames; the only planned code-touching changes (x5__ → mp__, currency/demo-value repin) are recorded as data-phase items.
• Personas × questions × endpoints (§6): 8 personas mapped to live API surface and the planned ops-layer surfaces.

Why

Plan step A1: the legend must be consistent across data, docs, and demo. This doc is the anchor the generator spec (A2), data phase (B), docs sweep (C1), and ops-layer design (D1) align to.

Verification

Docs-only change (one new file, no code, no nav/test coupling — mkdocs nav warns on omitted files only, no doc-sweeping tests match docs/*.md).

🤖 Generated with Claude Code

[github-actions]

DORA Metrics

• Window: last 30 days
• Branch: main
• Deployment frequency: 112 total / 26.13 per week
• Lead time for changes: avg 0.45h / median 0.0h
• Change failure rate: 75.89% (85/112)
• MTTR: 0.29h across 2 incident(s)