Direct Cloudflare deploys create Hyperdrive with query caching on #404

Problem

void deploy --platform cloudflare creates the Hyperdrive with the origin only. Cloudflare then uses its default query cache, so a read right after a write can return stale data for up to about 75 seconds.

void deploy --platform cloudflare
  └─ cf hyperdrive create --body { name, origin }      no caching setting
       └─ Cloudflare default: caching on (max_age 60 s + stale_while_revalidate 15 s)

app: INSERT ... ; SELECT ...   →  the SELECT can come from the cache

• Code: packages/void/src/cli/cloudflare-operation-process.ts (hyperdrive.create).
• Better Auth reads sessions through HYPERDRIVE (packages/void/src/runtime/better-auth-shared.ts), so a new session can look missing right after sign-in.
• The managed platform creates and updates project Hyperdrives with caching off, and turns it off for existing ones on deploy (fix(platform): address Cloudflare end-to-end deployment failures #399).

Questions

1. Should a Hyperdrive that Void creates on a direct deploy send caching: { disabled: true }?
2. Should Void change an existing Hyperdrive that it adopted by name?
3. A Hyperdrive that the user names in cloudflare.hyperdrive belongs to the user. Void should probably leave its settings alone. Confirm this.

Tests

• A direct deploy that creates a Hyperdrive sends caching off.
• A user-supplied cloudflare.hyperdrive id is not changed.