Module bgemm_naive 

Batched GEMM fallback using explicit loops. Naive batched GEMM kernel on strided views.

Operates on N-dimensional permuted views where dimensions are grouped as:

• A: [lo…, sum…, batch…]
• B: [sum…, ro…, batch…]
• C: [lo…, ro…, batch…]

Functions

bgemm_strided_into

Batched strided GEMM: C = alpha * A * B + beta * C

bgemm_strided_into_with_map

Batched strided GEMM with closure-based element mapping: C = alpha * map_a(A) * map_b(B) + beta * C