perf: make xtrain-cuda a regular dep of xtrain-optim (GPU AdamW)

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

crates/xtrain-optim/Cargo.toml

@@ -6,7 +6,5 @@ edition.workspace = true
 [dependencies]
 xtrain-tensor = { path = "../xtrain-tensor" }
 xtrain-autodiff = { path = "../xtrain-autodiff" }
-
-[dev-dependencies]
-# Acceptance tests drive the GPU (device selection) directly.
+# GPU AdamW (Phase T7) launches kernels + syncs the device directly.
 xtrain-cuda = { path = "../xtrain-cuda" }