feat(live): switch a MAINNET — micro-test fade-only €500 + funds-watch

Conversione testnet -> mainnet Deribit (soldi veri). Vedi
docs/specs/mainnet-microtest-plan.md, Fase 1.

- portfolios.yml: total_capital 500, leva 3, weighting equal; eseguono
  REALE solo i 7 single-leg (6 fade MR01/02/07 x BTC/ETH 15m + DIP01 1h);
  pairs/SH01/multi-asset -> paper (rumore arrotondamento a €500); real_truth.
- docker-compose.yml: portfolio+dashboard caricano anche .env.mainnet
  (token mainnet, prevale su .env; .env.mainnet resta gitignored).
- reconcile_account.py: watermark FONDI (compute_funds_change) — rileva
  aumenti di capitale (deposito/top-up) e cali anomali (prelievo) sul
  balance USDC, alert Telegram FUNDS_INCREASE/FUNDS_DECREASE; soglia
  max(€25, 5%). Stato in data/funds_watch.json (host-writable).
- .gitignore: ignora data/funds_watch.json (stato runtime).

Ledger testnet archiviato in data/_reset_backup/pre_mainnet_*.tgz e azzerato.
Crons host ri-puntati al conto reale (sourcing .env + .env.mainnet).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

docker-compose.yml

@@ -9,6 +9,7 @@ services:
       - ./portfolios.yml:/app/portfolios.yml:ro
     env_file:
       - .env
+      - .env.mainnet        # token MAINNET (soldi veri); prevale su .env (vedi micro-test plan)
     environment:
       - PYTHONUNBUFFERED=1
     healthcheck:
@@ -36,6 +37,7 @@ services:
       - ./docs/report:/app/docs/report:ro   # scheda strategie_attive.html (modal "scheda dettagliata")
     env_file:
       - .env
+      - .env.mainnet        # token MAINNET (soldi veri); prevale su .env (vedi micro-test plan)
     environment:
       - PYTHONUNBUFFERED=1
     healthcheck: