[release/2.7][ROCm][tunableop] UT tolerance increase for matmul_small_brute_force_… (#2397)

TunableOp will sometimes find a less precise solution due to the small
input vectors used in this UT. Bumping up tolerance to eliminate
flakiness.

Pull Request resolved: pytorch#158788
Approved by: https://github.com/jeffdaily

(cherry picked from commit c917c63)
(cherry picked from commit 35daec9)

Author: naromero77amd

test/test_linalg.py

@@ -4762,6 +4762,7 @@ def test_matmul_small_brute_force_3d_Nd(self, device, dtype):
     @onlyCUDA
     @skipCUDAIfNotRocm  # Skipping due to SM89 OOM in CI, UT doesn't do much on NV anyways
     @dtypes(*floating_types_and(torch.half))
+    @precisionOverride({torch.float16: 1e-1})  # TunableOp may occasionally find less precise solution
     def test_matmul_small_brute_force_tunableop(self, device, dtype):
         # disable tunableop buffer rotation for all tests everywhere, it can be slow
         # We set the TunableOp numerical check environment variable here because it is