Add rulebook RAG pipeline and LLM-driven game setup wizard

RAG: switch Postgres to pgvector, chunk and embed the three D&D
rulebooks locally via sentence-transformers, and retrieve relevant
excerpts per DM turn (query = latest player message) to ground the
system prompt. Retrieval runs off the event loop and is capped by a
relevance threshold and a max character budget so it can't blow up
context size or cost.

Game setup wizard: creating a game now opens a short chat where the
DM asks about genre, length, and the player's experience level, then
proposes a name and description via a tool call. The player can edit
both before creating the game. Stateless endpoint — the frontend
carries the conversation, no DB needed since the game doesn't exist
yet.

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

backend/app/rag/chunking.py

@@ -0,0 +1,33 @@
+import re
+
+# Rough word-count targets standing in for the 300-800 token / 10-20% overlap guideline —
+# English prose runs ~0.75 words per token, so ~380 words ≈ 500 tokens.
+WORDS_PER_CHUNK = 380
+OVERLAP_WORDS = 60
+
+_WHITESPACE_RE = re.compile(r"[ \t]+")
+_BLANK_LINES_RE = re.compile(r"\n{3,}")
+
+
+def _normalize(text: str) -> str:
+    text = _WHITESPACE_RE.sub(" ", text)
+    text = _BLANK_LINES_RE.sub("\n\n", text)
+    return text.strip()
+
+
+def chunk_text(
+    text: str, words_per_chunk: int = WORDS_PER_CHUNK, overlap_words: int = OVERLAP_WORDS
+) -> list[str]:
+    words = _normalize(text).split()
+    if not words:
+        return []
+
+    step = words_per_chunk - overlap_words
+    chunks = []
+    start = 0
+    while start < len(words):
+        chunks.append(" ".join(words[start : start + words_per_chunk]))
+        if start + words_per_chunk >= len(words):
+            break
+        start += step
+    return chunks