feat(quotes): anexo de conformidade ANATEL/RGC no PDF de oferta

Porta o anexo regulatório do legado (Resolução Anatel 765/2023, RGC,
recorte STFC) com fidelidade total ao conteúdo legal, incluindo a
seção de tarifação — em vez do html2canvas+jsPDF client-side do
legado, usa o Chromium server-side já construído para Documents.

Schema: products ganha service_type/metered/tariff_rates/has_ldi
(migration 1788525270647) — campos que a migration original de
products tinha deliberadamente deferido. Snapshot no item da oferta
na criação (nunca referência viva ao produto). Novo pacote
compartilhado packages/domain-shared/src/telecom para o catálogo de
tipos e legendas de tarifação.

Endpoint GET /v1/quotes/:id/anatel-pdf (400 sem item STFC), sempre
renderizado ao vivo (nunca cacheado, já que o conteúdo deriva de
updated_at). Frontend: campos de produto/oferta necessários +
"Baixar PDF ANATEL" na tela de detalhe da oferta. Inclui também o
botão de download de PDF em Documentos (endpoint já existia, faltava
o link na UI).

Verificado ao vivo através do container real: PDF gerado com
Chromium de verdade, texto extraído confere byte a byte com os
valores computados (multa, datas, tarifas, resumo financeiro).

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01BcWWuKfykGtUweXmHD6cKx

apps/core-web/src/lib/api/documents.ts

@@ -2,11 +2,12 @@
  * Client for `apps/api` `/v1/documents/templates` and `/v1/documents/generations`
  * — the "templates + generation" half of Documents (eden.md módulo 4, §1.1-1.3).
  *
- * This leva has NO real visual editor and NO real PDF: `editorContent` is a
- * simplified JSONB with a single root-level `plainText` string carrying
- * `{{variavel}}` placeholders (no rich node tree), and "generation" just
- * merges variables into `renderedHtml` — `pdf_s3_key`/`pdf_sha256`/
- * `pdf_size_bytes` are always `null` here. Don't build UI implying more.
+ * This leva has NO real visual editor: `editorContent` is a simplified
+ * JSONB with a single root-level `plainText` string carrying
+ * `{{variavel}}` placeholders (no rich node tree). PDF generation itself
+ * IS real (Chromium via packages/integrations/src/pdf/renderer.ts) —
+ * `pdf_s3_key`/`pdf_sha256`/`pdf_size_bytes` are populated on every
+ * `status: 'final'` generation.
  *
  * Like Contracts/Resellers, template/version/generation responses are
  * snake_case straight off the DB row — except the two computed fields on a
@@ -14,7 +15,7 @@
  * camelCase exactly as documented in the task brief; mirrored as-is below,
  * not "fixed" to snake_case.
  */
-import { apiRequest } from "./http.js";
+import { API_BASE_URL, apiRequest } from "./http.js";
 
 export type DocGroup = "cliente" | "revenda";
 export type DocumentVersionStatus = "draft" | "published";
@@ -184,6 +185,29 @@ export function createGeneration(
   );
 }
 
+/**
+ * Fetches the generated PDF and triggers a browser download. Not routed
+ * through `apiRequest` (JSON-only) — this needs the raw binary response,
+ * but still needs the bearer token, so a plain `<a href>` won't work
+ * either (the API route is behind JwtAuthGuard). Standard SPA pattern:
+ * fetch as a Blob, hand the browser a temporary object URL.
+ */
+export async function downloadGenerationPdf(id: string, accessToken: string, filename: string): Promise<void> {
+  const response = await fetch(`${API_BASE_URL}/documents/generations/${id}/pdf`, {
+    headers: { Authorization: `Bearer ${accessToken}` },
+  });
+  if (!response.ok) {
+    throw new Error("Não foi possível baixar o PDF.");
+  }
+  const blob = await response.blob();
+  const url = URL.createObjectURL(blob);
+  const link = document.createElement("a");
+  link.href = url;
+  link.download = filename;
+  link.click();
+  URL.revokeObjectURL(url);
+}
+
 /** Extracts `{{variavel}}` references from a plain-text template body — used to auto-build the generation context form. */
 export function extractTemplateVariables(plainText: string): string[] {
   const found = new Set<string>();