Add DataFrame entry points and replace apply_signal with signal_dates

[michaelchu]

Summary

• options_data(df) — validates/normalizes an existing DataFrame for use with strategies (no CSV or column-index mapping needed)
• load_cached_options(symbol) — reads options data from the parquet cache (produced by optopsy-data download) into the strategy pipeline
• load_cached_stocks(symbol) — reads cached stock OHLCV data for signal building
• signal_dates(stock_data, signal_func) — replaces apply_signal as the single way to generate entry/exit dates from signals

Removes apply_signal which confusingly required options data to build stock-based signals. signal_dates cleanly separates concerns: signals run on stock data, strategies run on options data.

New user flow

import optopsy as op

stocks = op.load_cached_stocks("SPY")
options = op.load_cached_options("SPY")

entry = op.signal_dates(stocks, op.rsi_below(threshold=30))
results = op.long_calls(options, entry_dates=entry)

Test plan

• All 1268 tests pass
• Pre-commit hooks pass (ruff format, ruff check, ty, pytest, branch name)
• Zero remaining references to apply_signal in codebase
• Docs and samples updated to show new flow

🤖 Generated with Claude Code