tests: fix some mul_mat test gaps (ggerganov#11375)

Now that we have batched mat-vec mul Vulkan shaders for up to n==8,
these tests weren't actually exercising the mat-mat mul path. Test
n==9 as well. Also, change to use all_types.

tests/test-backend-ops.cpp

@@ -3909,17 +3909,10 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
     test_cases.emplace_back(new test_gla(GGML_TYPE_F32, 32, 64, 32, 4));
     test_cases.emplace_back(new test_gla(GGML_TYPE_F32, 32, 64, 128, 4));
 
-    for (int i = 1; i < 9; ++i) {
-        test_cases.emplace_back(new test_mul_mat(GGML_TYPE_F16,    GGML_TYPE_F32, 16,  i, 256, { 1,  1}, {1, 1}));
-        test_cases.emplace_back(new test_mul_mat(GGML_TYPE_Q4_0,   GGML_TYPE_F32, 16,  i, 256, { 1,  1}, {1, 1}));
-        test_cases.emplace_back(new test_mul_mat(GGML_TYPE_Q4_1,   GGML_TYPE_F32, 16,  i, 256, { 1,  1}, {1, 1}));
-        test_cases.emplace_back(new test_mul_mat(GGML_TYPE_Q5_0,   GGML_TYPE_F32, 16,  i, 256, { 1,  1}, {1, 1}));
-        test_cases.emplace_back(new test_mul_mat(GGML_TYPE_Q5_1,   GGML_TYPE_F32, 16,  i, 256, { 1,  1}, {1, 1}));
-        test_cases.emplace_back(new test_mul_mat(GGML_TYPE_Q8_0,   GGML_TYPE_F32, 16,  i, 256, { 1,  1}, {1, 1}));
-        test_cases.emplace_back(new test_mul_mat(GGML_TYPE_Q4_K,   GGML_TYPE_F32, 16,  i, 256, { 1,  1}, {1, 1}));
-        test_cases.emplace_back(new test_mul_mat(GGML_TYPE_Q5_K,   GGML_TYPE_F32, 16,  i, 256, { 1,  1}, {1, 1}));
-        test_cases.emplace_back(new test_mul_mat(GGML_TYPE_Q6_K,   GGML_TYPE_F32, 16,  i, 256, { 1,  1}, {1, 1}));
-        test_cases.emplace_back(new test_mul_mat(GGML_TYPE_IQ4_NL, GGML_TYPE_F32, 16,  i, 256, { 1,  1}, {1, 1}));
+    for (ggml_type type_a : all_types) {
+        for (int i = 1; i < 10; ++i) {
+            test_cases.emplace_back(new test_mul_mat(type_a,    GGML_TYPE_F32, 16,  i, 256, { 1,  1}, {1, 1}));
+        }
     }
 
 #if 1