[tests] Fix gpu-to-cubin.mlir (NFC)

[anlunx]

Fix

[mehdi_amini]

FYI: I rolled this back, as well as the commit that introduced the bug, changing the pass pipeline is just hiding the issue here.

mlir/test/Integration/GPU/CUDA/gpu-to-cubin.mlir

 // RUN: mlir-opt %s \
 // RUN: | mlir-opt -gpu-kernel-outlining \
 // RUN: | mlir-opt -convert-vector-to-scf -convert-scf-to-cf -convert-cf-to-llvm -convert-vector-to-llvm -convert-arith-to-llvm \
 // RUN: | mlir-opt -pass-pipeline='builtin.module(gpu.module(strip-debuginfo,convert-gpu-to-nvvm,gpu-to-cubin))' \
-// RUN: | mlir-opt -gpu-to-llvm \
+// RUN: | mlir-opt -gpu-to-llvm -reconile-unrealized-casts \
 // RUN: | mlir-cpu-runner \
 // RUN:   --shared-libs=%mlir_cuda_runtime \
 // RUN:   --shared-libs=%mlir_runner_utils \
 // RUN:   --shared-libs=%mlir_c_runner_utils \
 // RUN:   --entry-point-result=void \
 // RUN: | FileCheck %s
 
 func.func @other_func(%arg0 : f32, %arg1 : memref<?xf32>) {