feat(rag): stream grounded RAG answers over SSE with source citations

Phase 03 (Story: Chat RAG Answer — happy path):
- app/rag/retriever.py: top-k cosine search + parent-doc selection with
  per-doc dedupe and BOR_MAX_CONTEXT_CHARS cap ([…truncated…] marker)
- app/rag/prompts.py: locked persona + HIGH/DEFLECT prompt builders
- app/rag/llm.py: LLMError + chat_stream (turbo, temp 0.4, max 700, stream)
- app/api/chat.py: POST /api/chat SSE — delta* then done{deflected,
  sources, suggestions}; query_log row + PLAN §9 per-turn log line;
  structured error event on mid-stream failure, JSON 503 when DB down
- frontend: SSE reader, live bubble streaming, source chips -> /sources.html,
  red role=alert banner, Send button state that always recovers
- fix(scaffold): [hidden] { display: none !important } — .kb-banner's
  display:flex was overriding the hidden attribute (banner always visible)
- tests: unit (retriever/prompts/sse/llm) + integration (real Postgres RAG
  turn, query_log, error + 503 paths, mid-turn failures) + Playwright story
  suite (grounded answer, log row, raw SSE shape); smoke placeholder test
  replaced with the real never-stale-button contract

tests/e2e/test_smoke.py

@@ -1,12 +1,10 @@
-"""Phase 01 smoke E2E: the app boots, serves the local frontend, and the
-placeholder chat round-trips without a stale button.
+"""Phase 01 smoke E2E: the app boots, serves the local frontend, and a chat
+round-trip never leaves a stale button (answer or error banner, both fine).
 
 Run:  uv run pytest tests/e2e/test_smoke.py -v
 """
 from __future__ import annotations
 
-import re
-
 import httpx
 from playwright.sync_api import Page, expect
 
@@ -27,19 +25,21 @@ def test_index_page_loads_locally(page: Page, app_url: str) -> None:
     assert 'href="http' not in html.replace('href="http://www.w3.org', "")
 
 
-def test_placeholder_chat_roundtrip(page: Page, app_url: str) -> None:
+def test_chat_roundtrip_never_stale_button(page: Page, app_url: str) -> None:
     page.goto(app_url)
     page.locator("#message-input").fill("hello brain")
     page.locator("#send-btn").click()
 
-    # User bubble appears, then the Brain placeholder answer arrives.
+    # User bubble appears first.
     page.locator(".msg.user .bubble").first.wait_for(state="visible", timeout=10_000)
-    brain_bubble = page.locator(".msg.brain .bubble").first
-    brain_bubble.wait_for(state="visible", timeout=10_000)
-    # to_have_text retries until the async fetch resolves (no stale read).
-    expect(brain_bubble).to_have_text(re.compile("neurons"), timeout=10_000)
+    # Then either a streamed Brain answer (DB up) or an error banner
+    # (DB down) — but the turn must always complete.
+    page.wait_for_selector(
+        ".msg.brain .bubble, #kb-banner.is-error",
+        state="visible",
+        timeout=20_000,
+    )
 
     # Button is never left stuck: back to "Send" and enabled.
-    btn = page.locator("#send-btn")
-    assert btn.is_enabled()
-    assert "Send" in btn.inner_text()
+    expect(page.locator("#send-btn")).to_be_enabled(timeout=10_000)
+    expect(page.locator("#send-label")).to_have_text("Send", timeout=10_000)