v1.5.0 — full MC/DCF/distribution parity (12 new tools)

Makes the Monte Carlo / DCF / distribution surface a faithful mirror of the okama engine. 33 → 45 tools.

New

• Custom distribution parameters — MCSpec.distribution_parameters lets you pin the forecast distribution (norm [mu, sigma], lognorm/t [shape|df, loc, scale]); any null element is fitted from history via MLE (e.g. a fixed Student-t df with auto loc/scale).
• MC distribution diagnostics — get_distribution_fit (fitted params, Jarque-Bera, Kolmogorov-Smirnov for the chosen + all distributions, backtesting error), get_return_moments (skewness/kurtosis, expanding or rolling), optimize_students_df (best-fit Student-t df vs empirical VaR/CVaR), get_cagr_distribution (simulated CAGR percentiles + probability below a score).
• DCF tools — get_dcf_wealth_index, get_dcf_cash_flow_ts, get_dcf_wealth_with_assets, get_survival_period (period + depletion date), get_initial_investment_values (PV/FV), get_monte_carlo_cash_flow (percentile bands).
• Charts — plot_qq (returns vs fitted distribution) and plot_hist_fit (histogram + fitted PDF).
• time_series_discounted_values flag on the time_series cash-flow strategy.

Changed

• All five CashFlow strategies remain available; discount_rate is now an optional argument on the PV/survival DCF tools.
• scipy is now a direct dependency.

Upgrade

uvx okama-mcp picks up the new version automatically; pinned installs: pip install -U okama-mcp.