Repoint account primary wallet so merge mail is delivered

[TaprootFreak]

Summary

Data-only migration that re-points one account's primary user_data.walletId from a wallet that disables the Info mail type to the default DFX Wallet (id 1, mailConfig = NULL).

This unblocks the account-merge confirmation mail for that account: ACCOUNT_MERGE_REQUEST is classified as MailContextType.INFO, and the account's primary wallet had Info in its mailConfig, so the mail was silently suppressed (mail.factory.ts isDisabledMailWallet → createMail returns undefined → notification.service.ts:26 sendMail returns without persisting). After re-pointing, resolveMailWallet (notification.service.ts:135) falls back to the default wallet (no Info opt-out) and the mail is delivered.

This PR is a targeted data fix only. The systemic bug (mandatory merge mail being suppressible because it lives in the Info bucket) is documented separately in #3842 and intentionally not fixed here.

Details

• New migration migration/1780993625289-RepointAccountWalletForMergeMail.js, hand-written in the style of the existing data migration 1779381590531-RouteScryptEurViaUsdt.js.
• Env-guarded: up() only acts when the row is in the expected pre-state (walletId = 67); it is a no-op on environments where the row/state differs. down() reverses it with a symmetric guard.
• PostgreSQL syntax (double-quoted identifiers); passes the migration-psql-check guard test.

Tradeoff

user_data.wallet also influences fee verification (fee.service.ts:201), auto-trade-approval (auth.service.ts:423) and mail branding. Re-pointing to the default wallet changes these bindings for this account — accepted as the intended outcome.

PR completeness

• Migration (data-only; no entity/column changes)
• Environment/Infrastructure — n/a
• Service updates — n/a (no DTO/interface changes)
• Frontend synchronization — n/a (no API contract changes)

Test plan

• npx jest migration-psql-check passes (PostgreSQL compatibility)
• npm run type-check passes
• Migration is env-guarded and idempotent; re-runs are no-ops

Relates to #3842