cascade-prior-v1

Submission-ready synthetic time-series generator for Cascade SN91.

It is intentionally pure NumPy and code-only. The generator mixes broad forecasting priors: trend, multi-seasonality, regime shifts, stochastic volatility, heavy-tail noise, intermittent demand, spikes, level shifts, bounded/saturating dynamics, and mean-reverting processes.

The output is deterministic for a fixed seed and yields finite univariate float64 arrays in the configured length band.