feat: per-league eligible sessions and break-even threshold settings (2.5.0)

Add two configurable leaderboard settings per league, replacing the global
constants with per-row DB columns:

- eligible_min_sessions (default 3): minimum sessions before a player ranks
  on the main leaderboard rather than showing as provisional
- break_even_cents (default 100 = $1.00): sessions within ±this of $0 net
  are classified as break-even rather than a win or loss

Changes:
- Migration 0003 adds both columns with server defaults for existing leagues
- League DB model gains eligible_min_sessions and break_even_cents columns
- net_result_bucket() and build_leaderboard() accept optional tolerance param
- summarize_player_runs() passes tolerance to net_result_bucket for streaks
- League leaderboard route reads values from the league row instead of config
- League settings form exposes both fields with validation (1-100 sessions,
  $0.00-$100.00 threshold)

services.py

@@ -103,7 +103,7 @@ def build_session_summaries(events: list[EventRow]) -> list[SessionSummary]:
     return sessions
 
 
-def summarize_player_runs(entries: list[SessionEntry]) -> dict[str, int | str | None]:
+def summarize_player_runs(entries: list[SessionEntry], break_even_cents: int = 100) -> dict[str, int | str | None]:
     ordered_entries = sorted(entries, key=entry_sort_key)
 
     longest_win_streak = 0
@@ -116,7 +116,7 @@ def summarize_player_runs(entries: list[SessionEntry]) -> dict[str, int | str |
 
     for entry in ordered_entries:
         net = entry.net_cents
-        bucket = net_result_bucket(net)
+        bucket = net_result_bucket(net, break_even_cents)
 
         if best_entry is None or net > best_entry.net_cents:
             best_entry = entry
@@ -141,7 +141,7 @@ def summarize_player_runs(entries: list[SessionEntry]) -> dict[str, int | str |
     current_loss_streak = 0
 
     for entry in reversed(ordered_entries):
-        bucket = net_result_bucket(entry.net_cents)
+        bucket = net_result_bucket(entry.net_cents, break_even_cents)
 
         if bucket == "win":
             if current_loss_streak > 0:
@@ -166,7 +166,7 @@ def summarize_player_runs(entries: list[SessionEntry]) -> dict[str, int | str |
     }
 
 
-def build_leaderboard(sessions: list[SessionSummary]) -> list[PlayerStats]:
+def build_leaderboard(sessions: list[SessionSummary], break_even_cents: int = 100) -> list[PlayerStats]:
     player_entries: dict[str, list[SessionEntry]] = defaultdict(list)
     for session in sessions:
         for entry in session.entries:
@@ -175,9 +175,9 @@ def build_leaderboard(sessions: list[SessionSummary]) -> list[PlayerStats]:
     leaderboard: list[PlayerStats] = []
     for player_name, entries in player_entries.items():
         nets = [entry.net_cents for entry in entries]
-        run_summary = summarize_player_runs(entries)
-        wins = [value for value in nets if net_result_bucket(value) == "win"]
-        losses = [value for value in nets if net_result_bucket(value) == "loss"]
+        run_summary = summarize_player_runs(entries, break_even_cents)
+        wins = [value for value in nets if net_result_bucket(value, break_even_cents) == "win"]
+        losses = [value for value in nets if net_result_bucket(value, break_even_cents) == "loss"]
 
         sessions_played = len(entries)
         winning_sessions = len(wins)