refactor(google-contacts-export): reduce mapToPerson cyclomatic complexity (currently 33)

[malek10xdev]

Context

backend/src/services/export/exports/google/contacts-api.ts:400 — mapToPerson has cyclomatic complexity 33 (DeepSource JS-R1005, "very-high" risk). The function is 130+ lines and handles 7+ different schema field types in one method.

The issue was flagged on PR #2830 (phone-only contacts) and intentionally not fixed in that PR because:

• It is pre-existing on main (function is 129 lines on main; we added only 5 lines for the hasEmail guard in commit fcbcccd).
• The refactor is non-trivial (~30-60 min) and carries regression risk without dedicated test coverage.
• It is out of scope for the phone-only contacts feature.

Proposed approach

Split mapToPerson into smaller helper functions, one per schema field:

• mapNames(contact, existing, updateEmptyOnly): Schema$Name[]
• mapEmails(contact, existing, updateEmptyOnly): Schema$EmailAddress[]
• mapPhones(contact, existing): Schema$PhoneNumber[]
• mapOrgs(contact, existing): Schema$Organization[]
• mapUrls(contact, existing): Schema$Url[]
• mapAddresses(contact, existing): Schema$Address[]
• mapMemberships(contact, existing, labels): Schema$Membership[]

Then mapToPerson becomes a ~10-line composer.

Existing test coverage

backend/test/unit/contact-export/googleContacts.test.ts:677 contains a describe('mapToPerson', ...) block that covers the happy path. May need additional branch coverage for the updateEmptyOnly flag and the per-field "is duplicate" checks.

Acceptance criteria

• mapToPerson cyclomatic complexity ≤ 20
• All existing googleContacts.test.ts tests pass
• No regression in Google Contacts sync UX (manual QA on a small export)
• DeepSource JS-R1005 no longer flags this file