[Upstream PyTorch] Deprecate `Fallback to CPU for XPU FP64` PyTorch patch

[vlad-penkin]

Check whether Fallback to CPU for XPU FP64 pytorch/pytorch#126516 patch is not needed anymore. If not deprecate the path.

Workflow	With fp64 patch	Without fp64 patch	Summary
Build and test GPU	all: passed: 16348, failed: 11, skipped: 4923, xfailed: 4164, total: 25446, fixme: 72 pass rate (w/o xfailed): 76.82%	all: passed: 16348, failed: 11, skipped: 4923, xfailed: 4164, total: 25446, fixme: 72 pass rate (w/o xfailed): 76.82%	pass rate equals
Build and test on Windows	all: passed: 16351, failed: 10, skipped: 4923, xfailed: 4164, total: 25448, fixme: 0 pass rate (w/o xfailed): 76.82%	all: passed: 16351, failed: 10, skipped: 4923, xfailed: 4164, total: 25448, fixme: 0 pass rate (w/o xfailed): 76.82%	pass rate equals
E2E performance			done (see comment below)
E2E accuracy			running
Triton benchmarks			hangs

[kwasd]

Patched:

• E2E accuracy main torchbench
• E2E performance main torchbench

Patch removed:

• E2E accuracy no 126516
• E2E performance no 126516

[kwasd]

WIP

• [Intel GPU] Enable fp64 GEMM pytorch/pytorch#140677

[anmyachev]

@kwasd pytorch/pytorch#140677 is merged

[kwasd]

• Triton benchmarks w/o fp64 patch https://github.com/intel/intel-xpu-backend-for-triton/actions/runs/13839553231
• E2E performance w/o fp64 patch https://github.com/intel/intel-xpu-backend-for-triton/actions/runs/13839546887
• E2E accuracy w/o fp64 patch https://github.com/intel/intel-xpu-backend-for-triton/actions/runs/13839534424

[anmyachev]

• Triton benchmarks w/o fp64 patch https://github.com/intel/intel-xpu-backend-for-triton/actions/runs/13839553231
• E2E performance w/o fp64 patch https://github.com/intel/intel-xpu-backend-for-triton/actions/runs/13839546887
• E2E accuracy w/o fp64 patch https://github.com/intel/intel-xpu-backend-for-triton/actions/runs/13839534424

@kwasd several workloads in "queued" status.