Add differentiable-SQL demos: ARCO-ERA5 and gradient descent

[alxmrs]

Stacked demo branch (on the autograd feature) holding the runnable benchmark scripts, kept out of the core branch so it stays reviewable.

• grad_era5.py: symbolic grad over real ARCO-ERA5 data (wind-speed sensitivity checked exactly; saturation vapour pressure checked against the closed-form Clausius-Clapeyron slope). The queries ORDER BY latitude DESC, longitude to match ERA5's native order, so results line up with the xarray reference with no sorting on either side (single partition, so the order survives to_dataset).
• grad_descent.py: gradient descent as ONE declarative recursive-CTE query, with grad() inside the recursion. The loss is written once and differentiated in place; AVG(grad(loss, a)) descends the gradient each step. No Python loop and no precompiled rule — grad() is rewritten to SQL before planning, so it works inside the recursive CTE. Fit matches numpy least-squares.