Pedagogy: add distribution parameter sliders to 20_probability notebooks

The `20_probability/` module currently has no interactive widgets. Distribution-parameter sliders are a natural fit — they're one of the textbook examples of "vary a parameter, see the shape change" that ipywidgets is designed for.

## Scope

Targets: `20_probability/10_probability.ipynb`, `20_probability/15_generating_random_numbers.ipynb`, `20_probability/20_statistics.ipynb`. Pick a subset for a first PR.

Per notebook: pick the central distribution covered and add sliders for its shape parameters, re-rendering the PDF (and CDF) as the user moves the sliders.

- **Normal**: μ, σ → bell curve; show how σ stretches and μ shifts.
- **Binomial**: n, p → discrete bars approaching normal as n grows (CLT preview).
- **Poisson**: λ → discrete bars; show approach to normal for large λ.
- **Exponential**: λ → memoryless decay.

Optional: empirical histogram from `numpy.random` overlaid on the analytic PDF, with a sample-size slider showing the histogram converging.

## Why

Distribution shape vs parameter is the foundational intuition of probability. Sliders carry it directly; tables don't. The empirical-overlay variant also previews the law of large numbers without naming it.

Validated slider pattern (#353/#354). The module has no widgets today, so this is a **whole-module uplift** rather than a notebook-by-notebook fix — one PR can probably cover all three target notebooks.