Document PlanetScale role successors

[klaemo]

Problem

PlanetScale Postgres branch roles can own database objects after an app or migration tool creates or changes schema objects. In branch-per-stage and preview workflows, SST removes non-production resources automatically. If a PostgresBranchRole owns objects and no successor is configured, PlanetScale can reject role deletion during preview cleanup because it does not know which role should receive ownership of those objects.

Representative cleanup failure from a preview stage:

~  Remove
|  Deleted     <dependent resources>
|  Error       <role-resource> planetscale:index:PostgresBranchRole
failure to invoke API: unknown status code returned: Status 422
{"code":"unprocessable","message":"Role is still referenced and cannot be dropped."}

✕  Failed

<role-resource> planetscale:index:PostgresBranchRole
failure to invoke API: unknown status code returned: Status 422
{"code":"unprocessable","message":"Role is still referenced and cannot be dropped."}

Fix

This updates the PlanetScale docs and Postgres examples to model successor: "postgres" on branch roles. The successor gives PlanetScale a role to reassign owned objects to before deleting a temporary branch role.

The docs now call out that runtime-only roles which never own objects do not require a successor, but setting one makes cleanup robust if the role later creates or owns objects.

Changes

• Add successor: "postgres" to the AWS PlanetScale Drizzle Postgres example.
• Add successor: "postgres" to the Cloudflare Hyperdrive PlanetScale example.
• Document why branch-per-stage Postgres roles should configure a successor.

Validation

• git diff --check
• bun run typecheck was attempted, but the current checkout fails on pre-existing platform AWS provider type errors before reaching these docs/example changes.