Add the missing _admonition/gpu.md shared include

Three lectures — autodiff, jax_intro and numpy_vs_numba_vs_jax — carry
`{include} _admonition/gpu.md`, but the file was never seeded into this
edition. Every other edition has it; this one did not:

    lecture-python-programming        lectures/_admonition/gpu.md
    lecture-python-programming.fa     lectures/_admonition/gpu.md
    lecture-python-programming.zh-cn  lectures/_admonition/gpu.md
    lecture-python-programming.fr     (absent)

So those three lectures have been building with a broken include and
publishing without the GPU admonition. The build reported success
throughout, because this repo built without -W.

Found by enabling -W in the same PR, which is the argument for the flag
in one example: three CRITICAL "file not found" directive errors that
had been live and invisible. This is the shared-asset class described in
QuantEcon/action-translation#117 — sync moves lecture files, so an asset
that lives outside that set is never carried across.

Translated per the French glossary (Lecture -> Cours) and typeset with
U+00A0 before the colon and inside the guillemets, matching the
convention applyTypography() enforces for prose directives — admonition
is on its PROSE_DIRECTIVES list, so the file is in scope for it.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: mmcky

lectures/_admonition/gpu.md

@@ -0,0 +1,12 @@
+```{admonition} GPU
+:class: warning
+
+Ce cours a été construit sur une machine disposant d'un GPU — bien qu'il fonctionne également sans.
+
+[Google Colab](https://colab.research.google.com/) propose une offre gratuite avec des GPU
+auxquels vous pouvez accéder comme suit :
+
+1. Cliquez sur l'icône « play » en haut à droite
+2. Sélectionnez Colab
+3. Réglez l'environnement d'exécution pour inclure un GPU
+```