implementing phase 3 pipeline: gemini structured output, validation, sqlite staging

Prompt is now a template (fixes the hardcoded en/es leftovers in rules
2, 3, 15, 16, 26, 31). pipeline.ts replaces the pseudocode: wordlist
normalization, skip-already-staged idempotency, batches of 20 against
gemini-3.6-flash with responseSchema, raw responses persisted per batch,
per-entry validation with rejection log, one transaction per word into
db/staging.db. Flags: --langs --pos --max-batches --delay-ms --dry-run.

Smoke run: 40/40 words staged (de+es, one batch each), 0 rejections.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

data-pipeline/tests/promptTemplate.test.ts

@@ -0,0 +1,57 @@
+import { describe, it, expect } from "vitest";
+import {
+  loadPromptTemplate,
+  renderPrompt,
+  type PromptParams,
+} from "../promptTemplate.js";
+
+const params: PromptParams = {
+  sourceLanguage: "es",
+  pos: "noun",
+  targetLanguages: ["en", "it", "de", "fr"],
+  words: ["suelo", "pared"],
+};
+
+describe("renderPrompt", () => {
+  it("substitutes every placeholder from the checked-in template", () => {
+    const rendered = renderPrompt(loadPromptTemplate(), params);
+    expect(rendered).toContain('Source language: Spanish ("es")');
+    expect(rendered).toContain('language must be "es"');
+    expect(rendered).toContain("Target languages: en, it, de, fr");
+    expect(rendered).toContain('"en" | "it" | "de" | "fr"');
+    expect(rendered).toContain("suelo\npared");
+    expect(rendered).toContain("valid Spanish noun");
+    expect(rendered).not.toMatch(/\{\{[A-Z_]+\}\}/);
+  });
+
+  it("does not tell the model to exclude a non-source language as target", () => {
+    const rendered = renderPrompt(loadPromptTemplate(), params);
+    expect(rendered).toContain(
+      'Do not include Spanish ("es") as a target_language.',
+    );
+    expect(rendered).not.toContain(
+      "Do not include English as a target_language",
+    );
+  });
+
+  it("throws when the source language is listed as a target", () => {
+    expect(() =>
+      renderPrompt(loadPromptTemplate(), {
+        ...params,
+        targetLanguages: ["en", "es"],
+      }),
+    ).toThrow(/exclude the source language/);
+  });
+
+  it("throws on an empty word batch", () => {
+    expect(() =>
+      renderPrompt(loadPromptTemplate(), { ...params, words: [] }),
+    ).toThrow(/empty word batch/);
+  });
+
+  it("throws on unreplaced placeholders", () => {
+    expect(() => renderPrompt("hello {{UNKNOWN_TOKEN}}", params)).toThrow(
+      /UNKNOWN_TOKEN/,
+    );
+  });
+});