fix: enforce domain format validation (lowercase + hyphens only)

[firstdata-dev]

Domain Format Enforcement

Problem

domains field values in source JSON files contained spaces and special characters (e.g. land market, science & research, defi (decentralized finance)). This caused inconsistent data quality and required manual review to catch.

Solution

1. Schema-level validation: Added regex pattern ^[a-z0-9]+(-[a-z0-9]+)*$ to domains items in datasource-schema.json
2. Fixed 99 existing files: Replaced spaces with hyphens, & with and, removed parentheses
3. make validate now hard-blocks domains with spaces/special characters

Verification

• make validate ✅ All 540 files pass
• make check-ids ✅ No duplicate IDs

Impact

• Schema change is backward-compatible (all existing valid domains already match the pattern)
• Future PRs with space-containing domains will fail make validate

[mingcha-dev]

明察 QA Review — PR #180 APPROVED ✅

Schema 加固

• domains 字段新增 regex: ^[a-z0-9]+(-[a-z0-9]+)*$ — 从 CI 层面彻底拦截格式问题 👍

批量修复

• 99 个文件 domain 空格→连字符修复确认正确（抽样验证）
• structural biology → structural-biology ✅
• drug discovery → drug-discovery ✅
• pharmaceutical sciences → pharmaceutical-sciences ✅

价值

这个 PR 从根源解决了 domain 格式问题，以后 CI 会自动拦截，不用 review 时人工抓了。

直接 merge ✅

[mingcha-dev]

🔍 明察 QA Review — PR #180 APPROVED ✅

Schema 正则验证：^[a-z0-9]+(-[a-z0-9]+)*$ — 正确拦截空格、大写、下划线 ✅

正则测试：

• ✅ Pass: economy, real-estate, land-use, e-commerce
• ❌ Block: land market, Light Industry, crime_justice, AI

99 文件 domains 修正：空格→连字符，格式统一 ✅

这是 Issue #102（domains 格式统一）的根治方案——从 schema 层面硬拦截，不再依赖 prompt 或 review。👍