feat(utils): enhance AEM Headless detection with SDK and GraphQL signals

[habansal]

Summary

Improves `detectAEMVersion()` in `spacecat-shared-utils` to detect AEM Headless sites more reliably.

Problem

1. `/content/dam/` is not headless-specific — CS and AMS sites reference DAM assets all the time; a CS site with images could be misclassified as headless
2. Single-signal detection was impossible — with only 2 patterns and `MIN_THRESHOLD=2`, both had to match; a site with just an `aem-headless` meta tag would fall through to `OTHER` and get waitlisted in PLG onboarding as "not an AEM site"

Changes

Replaced `/content/dam/` with purpose-built headless patterns:

Pattern	What it detects
`@adobe/aem-headless-client`	Official AEM Headless JS SDK (in bundled scripts)
`aem-headless-client`	Broader SDK reference
`graphql/execute.json`	Persisted GraphQL query calls
`content/_cq_graphql/`	AEM GraphQL endpoint references

Added weight boosting so each strong signal alone clears the detection threshold (consistent with how EDS/AMS/CS already use boosting):

• `aem-headless` string → +2 boost (3 pts total, clears threshold alone)
• `@adobe/aem-headless-client` → +3 boost (4 pts total)
• `graphql/execute.json` → +2 boost (3 pts total)
• `_cq_graphql` endpoint → +2 boost (3 pts total)

Tests

Replaced the two previous headless tests (both relied on having `aem-headless` AND `/content/dam/` together) with 5 targeted tests:

• ✅ `aem-headless` string alone → `AEM_HEADLESS`
• ✅ `@adobe/aem-headless-client-js` SDK reference → `AEM_HEADLESS`
• ✅ `/graphql/execute.json` persisted query → `AEM_HEADLESS`
• ✅ `/content/_cq_graphql/` endpoint → `AEM_HEADLESS`
• ✅ CS site with `/content/dam/` images → still `AEM_CS` (regression guard)

1155 tests passing, 0 failing.

Test plan

• Unit tests pass (`npm test -w packages/spacecat-shared-utils`)
• Lint passes (`npm run lint -w packages/spacecat-shared-utils`)

🤖 Generated with Claude Code