Add CONTD contract evaluator, deadline-aware buy pricing, actpkg JSON generator

- puga/contd.py, tools/contd.py: price open CONTD contract offers against real
  buy-fresh cost, flagging conditions that can't be priced from the API alone.
- tools/buyat.py: cheapest bid price likely to fill by a hard deadline
  (stockout risk), using the empirical first-passage model from history.py.
- puga/actpkg.py, tools/actpkg.py: Refined PrUn XIT ACT buy-list JSON packages.

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

tools/buyat.py

@@ -0,0 +1,83 @@
+#!/usr/bin/env python3
+"""Where to post a BUY order (a resting bid) when there's a hard DEADLINE - not "is patience
+worth it forever" (that's tools/sell.py's problem, where foregone margin is the only cost of
+waiting). Here, a bid that never fills means a production-line STOCKOUT (check tools/runway.py
+for your real deadline), so the right question is: what's the cheapest price that's still likely
+to fill BY that deadline, with paying the ask as the fallback if it doesn't?
+
+  puga buyat PG 150 --deadline 3        # 3 days of runway left on PG
+
+For each candidate price between the current ask and best bid, uses puga.history.empirical_touch
+with a fixed `horizon` (the deadline) to get the REAL empirical P(price actually falls to this
+level within that many days) - not "ever", which is the wrong question under a deadline. Expected
+cost = price * P(fill by deadline) + current_ask * (1 - P(fill by deadline)) - the fallback term
+assumes paying the ask at the deadline if the bid never filled; the ask itself could move by then,
+so this is an approximation, not a guarantee. Recommends the price minimizing expected cost, but
+shows the whole curve - pick a more conservative (higher-P) price than the cost-minimizer if a
+stockout would be expensive in ways this doesn't price in (e.g. the whole base's workforce idling).
+"""
+import argparse, sys
+from pathlib import Path
+sys.path.insert(0, str(Path(__file__).resolve().parent.parent))
+from puga import config, fio, market
+from puga.history import daily_prices, full_series, empirical_touch
+import math
+
+
+def main():
+    ap = argparse.ArgumentParser()
+    ap.add_argument("ticker")
+    ap.add_argument("qty", type=float)
+    ap.add_argument("--deadline", type=float, required=True, help="days until this input runs out (tools/runway.py) - the hard constraint, not a soft discount rate")
+    ap.add_argument("--cx", default=config.DEFAULT_CX)
+    a = ap.parse_args()
+    t = a.ticker.upper()
+    ob = fio.order_book(t, a.cx)
+    bids = sorted(((o["ItemCost"], o["ItemCount"] or 0) for o in ob["BuyingOrders"] if o.get("ItemCost")), reverse=True)
+    best_bid = bids[0][0] if bids else None
+    best_ask = ob.get("Ask")
+
+    hit = market.walk(t, a.cx, a.qty, "buy")
+    ask_cost = hit["total"] if not hit["short"] else None
+    print(f"{t}.{a.cx}  best bid {best_bid}  ask {best_ask}  vwap7 {ob.get('PriceAverage')}  deadline {a.deadline:g}d")
+    if ask_cost is not None:
+        print(f"instant (pay the asks now): {ask_cost:,.0f}  (avg {hit['avg']:.0f}) - the fallback if nothing fills in time")
+    else:
+        print("WARNING: book too thin to fill this quantity at any price right now - the 'pay the ask' fallback isn't actually available at this size")
+
+    dp = daily_prices(t, a.cx)
+    s0 = sorted(dp.items())[-1][1][0] if dp else (best_bid or ob.get("Bid") or 0)
+    series = full_series(t, a.cx)
+    if len(series) < 10:
+        print("\nnot enough price history to estimate fill probability - fall back to paying the ask")
+        return
+
+    candidates = sorted({p for p, _ in bids if p <= s0} | {round(s0 * m, 0) for m in (0.99, 0.97, 0.95, 0.9, 0.85, 0.8, 0.7, 0.6)})
+    candidates = [p for p in candidates if p > 0 and (best_bid is None or p >= best_bid * 0.5)]
+
+    print(f"\nlast traded {s0:.0f}, {len(series)}d history - P(fill) is specifically BY the {a.deadline:g}d deadline, not ever:")
+    print(f"  {'price':>10} {'P(fill by deadline)':>20} {'cost if filled':>15} {'expected cost':>14}")
+    rows = []
+    for p in candidates:
+        y = math.log(p / s0) if s0 else 0.0
+        if y >= 0:
+            p_fill = 1.0
+        else:
+            p_fill = empirical_touch(series, y, 0.0, horizon=a.deadline)["touched_frac"]
+        cost_if_filled = p * a.qty
+        fallback = ask_cost if ask_cost is not None else p * a.qty * 1.5   # crude: no real fallback price known
+        expected = p_fill * cost_if_filled + (1 - p_fill) * fallback
+        rows.append((p, p_fill, cost_if_filled, expected))
+
+    best = min(rows, key=lambda r: r[3]) if rows else None
+    for p, p_fill, cost_if_filled, expected in rows:
+        flag = "  <- min expected cost" if (p, p_fill, cost_if_filled, expected) == best else ""
+        print(f"  {p:>10.0f} {100*p_fill:>19.0f}% {cost_if_filled:>15,.0f} {expected:>14,.0f}{flag}")
+
+    print("\nNOTE: 'expected cost' assumes paying today's ask as the fallback if the bid never fills by the "
+          "deadline - the real ask could move by then, this is an estimate. Pick a price with HIGHER P(fill) "
+          "than the cost-minimizer if a stockout costs more than this model sees (idle workforce, missed orders).")
+
+
+if __name__ == "__main__":
+    main()