[BACKEND] Refine v100 tests and fix mmav1 numwarps>1 hang issue (#971)

This PR

- Fix numWarps>1 hang issue
- add existing test cases in test_gemm.py to CI, and add a common flag
`valid_on_Volta` to determine whether the test case should be activated
on Volta or just skip.
  - Currently, the column-major cases are disabled.
 - Add test_core.py and other tests to Volta CI
   - the `test_printf.py` failed.

lib/Dialect/TritonGPU/Transforms/Combine.cpp

@@ -617,13 +617,15 @@ SmallVector<unsigned, 2> warpsPerTileV1(triton::DotOp dotOp,
   bool changed = false;
   do {
     changed = false;
+    int pre = ret[0];
     if (ret[0] * ret[1] < numWarps) {
       ret[0] = std::clamp<unsigned>(ret[0] * 2, 1, shape[0] / shapePerWarp[0]);
-      changed = true;
+      changed = pre != ret[0];
     }
     if (ret[0] * ret[1] < numWarps) {
+      pre = ret[1];
       ret[1] = std::clamp<unsigned>(ret[1] * 2, 1, shape[1] / shapePerWarp[1]);
-      changed = true;
+      changed = pre != ret[1];
     }
   } while (changed);
   return ret;