feat(cpp-object-oriented-modular-refactoring): step 16 - generic-sparse-assembler

include/fesa/assembly/sparse_assembler.h

@@ -2,6 +2,7 @@
 #define FESA_ASSEMBLY_SPARSE_ASSEMBLER_H_
 
 #include "fesa/core/status.h"
+#include "fesa/elements/element.h"
 #include "fesa/math/sparse_matrix.h"
 
 namespace fesa {
@@ -13,10 +14,22 @@ class ParallelFor;
 /// @brief Owns deterministic element-contribution reduction into global CSR.
 class SparseAssembler {
  public:
+  /// @brief Assembles runtime element stiffness into validated full-DOF CSR.
+  /// @param elements Non-owning elements in stable active source order.
+  /// @param dofs Owner of the matching scatter and structural pattern.
+  /// @param parallel_for Backend for index-owned element-local computation.
+  /// @return A validated matrix or a structured model failure.
+  /// @note Workers write only their element-owned COO buffers; flattening and
+  /// duplicate reduction retain fixed element and local-entry order.
+  static Result<SparseMatrix> Assemble(const ElementView& elements,
+                                       const DofManager& dofs,
+                                       const ParallelFor& parallel_for);
+
   /// @brief Assembles stiffness in stable source-element and local-entry order.
   /// @return A validated full-DOF CSR matrix or structured model failure.
   /// @note Parallel workers produce index-owned local buffers; serial reduction
-  /// remains the sole global CSR writer.
+  /// remains the sole global CSR writer. This compatibility facade creates
+  /// runtime candidates until LinearStaticAnalysis owns them directly.
   static Result<SparseMatrix> AssembleStiffness(
       const AnalysisModel& model, const DofManager& dofs,
       const ParallelFor& parallel_for);