feat: Add allSiteIdsByTier and allByEnrollmentAndTier collection meth…

[sandsinh]

What

Adds two new collection methods to spacecat-shared-data-access for querying sites by entitlement tier:

• SiteEnrollmentCollection.allSiteIdsByTier(tier, productCode?) — single JOIN query against site_enrollments/entitlements returning matching siteIds, optionally narrowed to a product code.
• SiteCollection.allByEnrollmentAndTier(tier, productCode?, options?) — chains through SiteEnrollmentCollection via entityRegistry and batch-fetches full Site objects for those IDs.

Why

Callers need to look up sites by entitlement tier (e.g. PAID, FREE_TRIAL, PLG), optionally scoped to a product code (e.g. LLMO), without fetching every enrollment and filtering in memory.

Testing

Unit and integration tests added for both new methods (site.collection.test.js, site.test.js, site-enrollment.collection.test.js, site-enrollment.test.js).

Related Issues

SITES-43769

[github-actions]

This PR will trigger a minor release when merged.

[MysticatBot]

Hey @sandsinh,

Verdict: Request changes - one deduplication gap to address before merge.
Complexity: MEDIUM - medium diff, new collection methods following existing patterns.
Changes: Adds tier-based site enrollment lookup methods (allSiteIdsByTier and allByEnrollmentAndTier) with full test coverage (8 files).

Must fix before merge

1. [Important] Duplicate site IDs not deduplicated before batch fetch - packages/spacecat-shared-data-access/src/models/site-enrollment/site-enrollment.collection.js:79 (details inline)

Non-blocking (2): minor issues and suggestions

• nit: Inconsistent input validation - allSiteIdsByTier uses if (!tier) while allByEnrollmentAndTier uses if (!hasText(tier)). The falsy check accepts whitespace-only strings; consider aligning to hasText for consistency - site-enrollment.collection.js:59
• nit: TypeScript declarations for allSiteIdsByProductCode and allByEnrollmentProductCode are added to the .d.ts files but their implementations are not visible in this diff. Confirm these are pre-existing methods getting their missing types backfilled alongside the new code - site-enrollment/index.d.ts:9, site/index.d.ts:66

---

_{Skill: pr-review | Model: us.anthropic.claude-opus-4-6-v1[1m] | Duration: 2m 41s | Cost: $4.78 | Commit: 71632b2245ae3b16aa6e6f6926c90ebf477b7b20
If this code review was useful, please react with 👍. Otherwise, react with 👎.}

[MysticatBot]

issue (blocking): allSiteIdsByTier returns raw (data || []).map((row) => row.site_id) without deduplication. If a site is enrolled in multiple entitlements at the same tier (e.g., two PAID entitlements for different product codes when no productCode filter is supplied), the query returns duplicate site_id values. These duplicates flow into allByEnrollmentAndTier and cause redundant batchGetByKeys lookups - wasted I/O on the same primary keys.

The existing allSiteIdsByProductCode sibling has the same pattern but is less likely to produce duplicates in practice (unique on (site_id, entitlement_id) plus a specific product code narrows to one entitlement per site). Tier is a broader classifier and more likely to hit this.

Fix:

return [...new Set((data || []).map((row) => row.site_id))];