test-backend-ops: add Bonsai-shape MUL_MAT cases for PTQ1_0/PQ2_0

The stock generator only emits 16 x 256 cases for these types, all of
which take the single-column mmvq path, so the multi-column dispatchers
(ncols_dst 2..8) had no coverage at all.

Add the real Bonsai row lengths (k = 1024/5120/6144/17408) at odd row
counts (m = 67/70, landing a partial row group) with n sweeping 1..8.

Arc Pro B50, oneAPI 2026.1, Level Zero, per type:
  MUL_MAT     9 -> 102/102 passed
  MUL_MAT_ID       75/75 passed

Author: kiljoy001

tests/test-backend-ops.cpp

@@ -9301,6 +9301,19 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
     test_cases.emplace_back(new test_mul_mat(GGML_TYPE_Q8_0, GGML_TYPE_F32, 2880, 32, 2880, {1, 1}, {1, 1}));
     test_cases.emplace_back(new test_mul_mat(GGML_TYPE_MXFP4, GGML_TYPE_F32, 2880, 32, 2880, {1, 1}, {1, 1}));
 
+    // Ternary formats at real Bonsai weight shapes. The k values are the model's actual row
+    // lengths and the odd m values land a partial row group; n sweeps 1..8 so the multi-column
+    // mmvq dispatchers run, which the 16 x 256 cases above never reach.
+    for (ggml_type type_a : {GGML_TYPE_PTQ1_0, GGML_TYPE_PQ2_0}) {
+        for (int64_t k : {1024, 5120, 6144, 17408}) {
+            for (int64_t m : {67, 70}) {
+                for (int64_t n = 1; n <= 8; ++n) {
+                    test_cases.emplace_back(new test_mul_mat(type_a, GGML_TYPE_F32, m, n, k, {1, 1}, {1, 1}));
+                }
+            }
+        }
+    }
+
     // m == 1, with n on both sides of MMVF_MAX_BATCH_SIZE (8): mmvf below, operand swap above
     for (int64_t n : {1, 7, 8, 9, 16, 128, 512}) {
         test_cases.emplace_back(new test_mul_mat(GGML_TYPE_F32, GGML_TYPE_F32, 1, n, 2048, {1, 1}, {1, 1}));