Games: in-progress hub status + distribution-aware word-search placement (Codex)

- Play hub: word cards now surface IN-PROGRESS games too (not just won/lost) so
  "continue on another device" shows at a glance — card reads "5:3…" and the
  selection option says "Continue · 3/6".
- Word Search generator: replace "prefer any crossing" with a SCORED placement —
  score = overlap*4 - local crowding (filled neighbours that aren't crossings) —
  then pick among the best ~20%. Keeps the organic interlocking but spreads words
  across the board instead of clumping around the first-placed (longest) words.
  Every word still placed (tests green). NOTE: changes today's grid layouts, so
  an in-progress word search resets once.

237 pytest + 11 vitest green.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

goodnews/games.py

@@ -309,6 +309,20 @@ def adjudicate_word_guess(conn: sqlite3.Connection, date: str, variant: str, gue
 
 _DIRS = [(-1, -1), (-1, 0), (-1, 1), (0, -1), (0, 1), (1, -1), (1, 0), (1, 1)]
 
+
+def _neighbour_fill(grid, cells, size: int) -> int:
+    """Filled cells in a candidate's footprint+border that AREN'T its own cells —
+    a crowding measure, so placement can spread words out instead of clumping."""
+    own = set(cells)
+    rs = [r for r, _ in cells]
+    cs = [c for _, c in cells]
+    cnt = 0
+    for r in range(max(0, min(rs) - 1), min(size, max(rs) + 2)):
+        for c in range(max(0, min(cs) - 1), min(size, max(cs) + 2)):
+            if (r, c) not in own and grid[r][c] is not None:
+                cnt += 1
+    return cnt
+
 # Size tiers. The three sizes draw DISJOINT word slices from the day's pool, so
 # each is its own fresh puzzle (no repeats across sizes). small+med+large counts
 # sum to WS_NEEDED, the minimum unique words a theme must supply.
@@ -513,23 +527,25 @@ def _build_grid(words: list[str], size: int, seed: int) -> tuple[list[str], list
     for word in sorted(words, key=len, reverse=True):
         if len(word) > size:
             continue
-        # Gather valid placements, then PREFER one that crosses an already-placed
-        # word (shares a matching letter) so the grid interlocks like a real word
-        # search — falling back to any valid spot so every word still gets placed.
-        valid = []  # (overlap_count, cells)
+        # Gather valid placements and SCORE them: reward crossing an existing word
+        # (so the grid interlocks like a real puzzle) but penalise crowding, so
+        # words spread across the board instead of all clustering around the ones
+        # placed first. Pick at random among the best ~20% to keep organic variety.
+        scored = []  # (score, cells)
         for _ in range(400):
             dr, dc = rng.choice(_DIRS)
             r0, c0 = rng.randrange(size), rng.randrange(size)
             cells = [(r0 + dr * i, c0 + dc * i) for i in range(len(word))]
             if any(not (0 <= r < size and 0 <= c < size) for r, c in cells):
                 continue
-            if all(grid[r][c] in (None, word[i]) for i, (r, c) in enumerate(cells)):
-                overlap = sum(1 for i, (r, c) in enumerate(cells) if grid[r][c] == word[i])
-                valid.append((overlap, cells))
-        if not valid:
+            if not all(grid[r][c] in (None, word[i]) for i, (r, c) in enumerate(cells)):
+                continue
+            overlap = sum(1 for i, (r, c) in enumerate(cells) if grid[r][c] == word[i])
+            scored.append((overlap * 4 - _neighbour_fill(grid, cells, size), cells))
+        if not scored:
             continue
-        crossing = [c for c in valid if c[0] > 0]
-        _, cells = rng.choice(crossing if crossing else valid)
+        scored.sort(key=lambda t: t[0], reverse=True)
+        _, cells = rng.choice(scored[: max(1, len(scored) // 5)])
         for i, (r, c) in enumerate(cells):
             grid[r][c] = word[i]
         placed.append(word)