[MLIR] XeVM `gpu-lower-to-xevm-pipeline` leaves scalar `math.isinf` unconverted

On LLVM main, the XeVM lowering pipeline leaves a scalar `math.isinf` operation inside an otherwise LLVM-lowered `gpu.module`. Kernel serialization then fails because the Math dialect operation cannot be translated to LLVM IR.

### Input Program
`input.mlir`
```llvm
module attributes {gpu.container_module} {
gpu.module @kernels {
gpu.func @classify_inf(%input : memref<1xf32>,
%output : memref<1xi32>) kernel {
%c0 = arith.constant 0 : index
%zero = arith.constant 0 : i32
%one = arith.constant 1 : i32
%value = memref.load %input[%c0] : memref<1xf32>
%is_inf = math.isinf %value : f32
%result = arith.select %is_inf, %one, %zero : i32
memref.store %result, %output[%c0] : memref<1xi32>
gpu.return
}
}
}
```

### Reproduction Command
```bash
mlir-opt input.mlir '-gpu-lower-to-xevm-pipeline=xegpu-op-level=lane zebin-triple=spirv64-unknown-unknown zebin-chip=bmg opt-level=2 binary-format=fatbin'
```

The command fails with:
```bash
input.mlir:9:17: error: cannot be converted to LLVM IR: missing
LLVMTranslationDialectInterface registration for dialect for op: math.isinf
%is_inf = math.isinf %value : f32
^
input.mlir:9:17: note: see current operation:
%3 = "math.isinf"(%2) ... : (f32) -> i1
input.mlir:2:3: error: Failed creating the llvm::Module.
input.mlir:2:3: error: An error happened while serializing the module.
```

Version: 13e5ce40eff159ed9b2db3660b41c47ca640ce51.

Start with the input.mlir reproducer and run the documented mlir-opt gpu-lower-to-xevm-pipeline command to confirm the failure. Trace the XeVM lowering pipeline handling of the scalar math.isinf operation; done means the operation is converted or otherwise handled so kernel serialization succeeds without the missing LLVMTranslationDialectInterface error.

Written by the indexing model from the issue text.