fix(d1_database): add read_replication default to prevent v5 apply failure

[ssicard]

Summary

Add read_replication = { mode = "disabled" } to every cloudflare_d1_database resource block during v4→v5 migration.

Problem

The E2E CI run (#278) is failing because all cloudflare_d1_database resources get a 400 Bad Request from the D1 API:

"Invalid property: read_replication => Expected object, received null"

Root cause chain:

1. v4 state has no read_replication field (it did not exist in v4)
2. tf-migrate runs — the d1_database migrator was a no-op, so no read_replication is added to the config
3. The v5.22.0 provider's new state upgrader (UpgradeFromLegacyV0) initializes read_replication as nil in state
4. Terraform sends read_replication: null in the PUT request
5. The D1 API rejects null — it expects {"mode": "disabled"}, not null

Fix

TransformConfig now sets read_replication = { mode = "disabled" } on every d1_database block that does not already have it. This matches the API default and prevents the null-rejection error.

A corresponding provider-side fix (initializing read_replication to {mode:"disabled"} in the state upgrader) will be submitted separately to terraform-provider-cloudflare.

Changes

• internal/resources/d1_database/v4_to_v5.go — set read_replication default in TransformConfig
• internal/resources/d1_database/v4_to_v5_test.go — updated 5 expected outputs + added idempotency test
• integration/v4_to_v5/testdata/d1_database/expected/d1_database.tf — all 8 resource blocks updated

Testing

go test -v ./internal/resources/d1_database/...   # 6/6 pass
TEST_RESOURCE=d1_database go test -v -run TestSingleResource ./integration/...  # pass