Skip fake autocast tests for `linalg.pinv` and `pinverse` on XPU (#178277)

Add `fake_autocast_device_skips` entry for XPU, aligned with the existing CUDA and CPU behavior. FakeTensor cannot correctly model autocast dtype promotion through `at::matmul` (called internally by `linalg_pinv` and `pinverse`), producing float32 output while real autocast produces float16.

CUDA already skips these tests (added in [1]). The same root cause applies to XPU since it shares the same autocast operation lists and default dtype.

[1] https://github.com/pytorch/pytorch/pull/114531
Pull Request resolved: https://github.com/pytorch/pytorch/pull/178277
Approved by: https://github.com/tszulist-hbn, https://github.com/kdrozd-dev, https://github.com/guangyey, https://github.com/jansel