Update PyTorch pin upto b36a20d (#4965)

*
https://github.com/intel/intel-xpu-backend-for-triton/actions/runs/17264859706
(FlexAttn - failed:
`inductor/test_flex_attention.py::TestFlexAttentionXPU::test_non_contiguous_last_dim_xpu
- AssertionError: False is not true : Out Compiled error
0.4045442892146799 is greater than ref error 0.00021200643481965935 by
more than 10.0X.`)
*
https://github.com/intel/intel-xpu-backend-for-triton/actions/runs/17264866621
(FlexDec - passed)

Seems #4927
(FlexDec) is fixed but we have one new failure for FlexAttn

Track issue:
#4968

Signed-off-by: Anatoly Myachev <[email protected]>

Author: anmyachev

.github/pins/pytorch.txt

@@ -1 +1 @@
-818ba434c7de4cd604184b2857d544e0ad95735f
+b36a20d368733740a8507b3109d193c88930323a