fix: install PaymentMethod CRDs + restore webhook Service name

[mattdjenkinson]

Summary

Two release-process bugs that landed in v0.2.0 and still exist on main — together they take down billing-controller-manager and break BillingAccount admission against the deployed bundle in staging.

Bug 1 — PaymentMethod / PaymentMethodClass CRDs in the bundle but never applied

`config/base/crd/kustomization.yaml` only lists four CRDs even though controller-gen had emitted six. `PaymentMethod` and `PaymentMethodClass` exist in `config/base/crd/bases/` but were never referenced, so the bundle that ships to Flux installs only the four originals.

Live impact in staging:

```
ERROR controller-runtime.source.Kind if kind is a CRD, it should be installed before calling Start
{"kind": "PaymentMethod.billing.miloapis.com", "error": "no matches for kind "PaymentMethod" in version "billing.miloapis.com/v1alpha1""}
ERROR Could not wait for Cache to sync ... timed out waiting for cache to be synced for Kind *v1alpha1.PaymentMethod
```

`billing-controller-manager` exits 1, CrashLoopBackOff, 13 restarts on each replica.

Bug 2 — Webhook Service name regression

`config/base/webhook/manifests.yaml` ships with the kubebuilder placeholder Service name (`webhook-service`) instead of the `billing-webhook` Service this repo actually declares. Kustomize's `nameReference` transformer cannot bridge the two because nothing ever renames the Service — they are simply different identifiers.

The deployed `MutatingWebhookConfiguration` therefore points at `webhook-service.billing-system.svc`, which does not exist, and every `BillingAccount` mutating-admission call fails DNS lookup:

```
BillingAccount/milo-system/fake-usage dry-run failed (InternalError):
Internal error occurred: failed calling webhook "mbillingaccount.kb.io":
failed to call webhook: Post "https://webhook-service.billing-system.svc:443/mutate-billing-miloapis-com-v1alpha1-billingaccount?timeout=10s":
dial tcp: lookup webhook-service.billing-system.svc on 10.2.0.10:53: no such host
```

The existing remediation was a comment in `webhook/kustomization.yaml` asking devs to hand-restore the service name after each `task manifests` regeneration. v0.2.0 demonstrates how reliable that is.

Fix

File	Change
`config/base/crd/kustomization.yaml`	Add the two missing CRD entries.
`config/base/webhook/manifests.yaml`	Restore the eight `name: webhook-service` references to `name: billing-webhook`.
`Taskfile.yaml` (`manifests` task)	Append an in-place `perl -pi` step so the next regeneration is idempotent. The footgun comment in `webhook/kustomization.yaml` is replaced with one that documents the new automated fix-up.

Validation

• `kustomize build config/base/crd` lists all six CRDs.
• `kustomize build config/base/webhook` clean.
• Re-ran `task manifests` end-to-end: `grep -c "name: billing-webhook"` → 8, `grep -c "name: webhook-service"` → 0 in the regenerated file.

Deployment recovery

Once this lands and a new bundle tag publishes, FluxCD will reconcile `billing-milo-control-plane` (CRDs land), `billing-milo-webhook` (MWC Service ref corrected), and `billing-manager` (Deployment becomes healthy once its CRD watches succeed).