GPU: Add BLAS-1 ops, DeviceScalar, device-resident SpMV, and CG interop (5/5)

Add the operator interface needed for GPU iterative solvers:

- BLAS Level-1 on DeviceMatrix: dot(), norm(), squaredNorm(), setZero(),
  noalias(), operator+=/-=/\*= dispatching to cuBLAS axpy/scal/dot/nrm2.
- DeviceScalar<Scalar>: device-resident scalar returned by reductions.
  Defers host sync until value is read (implicit conversion). Device-side
  division via NPP for real types.
- GpuContext: stream-borrowing constructor, setThreadLocal(), cublasLtHandle(),
  cusparseHandle().
- GEMM upgraded from cublasGemmEx to cublasLtMatmul with heuristic algorithm
  selection and plan caching.
- GpuSparseContext: GpuContext& constructor for same-stream execution,
  deviceView() returning DeviceSparseView with operator* for device-resident
  SpMV (d_y = d_A * d_x).
- geam expressions: d_C = d_A + alpha * d_B via cublasXgeam.
- GpuSVD::matrixV() convenience wrapper.

These additions make DeviceMatrix usable as a VectorType in Eigen algorithm
templates. Conjugate gradient is the motivating example and is tested against
CPU ConjugateGradient for correctness.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Eigen/GPU

@@ -39,6 +39,7 @@
 
 #ifdef EIGEN_USE_GPU
 // IWYU pragma: begin_exports
+#include "src/GPU/DeviceScalar.h"
 #include "src/GPU/DeviceMatrix.h"
 #include "src/GPU/GpuContext.h"
 #include "src/GPU/DeviceExpr.h"
@@ -53,8 +54,10 @@
 #include "src/GPU/CuFftSupport.h"
 #include "src/GPU/GpuFFT.h"
 #include "src/GPU/CuSparseSupport.h"
+#ifdef EIGEN_SPARSECORE_MODULE_H
 #include "src/GPU/GpuSparseContext.h"
-#ifdef EIGEN_CUDSS
+#endif
+#if defined(EIGEN_CUDSS) && defined(EIGEN_SPARSECORE_MODULE_H)
 #include "src/GPU/CuDssSupport.h"
 #include "src/GPU/GpuSparseSolverBase.h"
 #include "src/GPU/GpuSparseLLT.h"