feat: add reusable performance report helpers

[konradbachusz]

Lift the strategy-agnostic pieces of get_performance.py from the algorithmic-trading-project repo into this library so any strategy can produce snapshot + PDF performance reports with one call.

New public API:

• brokers.alpaca.performance_ops.get_portfolio_equity_series() Wraps get_portfolio_history() into a date-indexed pd.Series.

• common.portfolio_ops.fetch_normalized_benchmark(portfolio_equity, date_start) Fetches S&P 500, intersects to common dates, and rescales the benchmark so its first value equals the portfolio's first value (units mismatch between index points and dollars).

• common.viz_ops.build_performance_figures(viz, show, mask_benchmark_on) Expands create_all_plots() with per-plot benchmark masking. Defaults to hiding the benchmark line on the dollar-scale plots (cumulative_returns, equity_curve) where the raw S&P index has no meaningful comparison.

• common.report_ops.write_performance_pdf(pdf_path, title, metrics, figs, period_text) Multi-page PDF: cover with monospace metrics block + one chart per page.

• brokers.alpaca.performance_ops.generate_performance_report(...) End-to-end orchestrator returning (snapshot_path, pdf_path, metrics).

Tests:

• New tests/test_report_ops.py covering PDF cover + figure pages, empty metrics, zero figures, and string-vs-Path inputs.
• Tests for fetch_normalized_benchmark covering normalisation, common-date intersection, and date_start forwarding.
• Tests for build_performance_figures covering default masking, restoration of viz.benchmark, full-benchmark mode, and alpha/beta skipping.
• Tests for get_portfolio_equity_series and generate_performance_report (snapshot + PDF + metrics; benchmark-disabled path).

All 202 tests pass (sentiment_ops collection error pre-existing on main).