Diffusion / flow specs (umbrella)

This is the curated public entrypoint for TorchLean's diffusion / flow spec layer.

It re-exports:

• a discrete VP schedule + forward noising (qSample),
• reverse samplers (DDPM, deterministic DDIM, and the clipped image sampler), and
• a continuous-time VP schedule + probability-flow ODE drift (pfOdeRhs).

All specs are scalar-polymorphic (Context α) so the same definitions can be reused for:

• runtime execution (Float, Float32, or the configured binary32 type ExecFloat.Binary 8 23),
• CPU software execution at a chosen precision (FloatLib.Floats.ExecFloat.Binary),
• proofs (ℝ or the noncomputable rounded-real model FloatLib.Floats.Formats.Flocq.NF), and
• verification backends (interval scalars, etc.).