[CPU] Remove outdated llvm::cl::opt cmd flag. (iree-org#17528)

The flag is not used at all.

compiler/src/iree/compiler/Codegen/LLVMCPU/Passes.cpp

@@ -54,15 +54,6 @@ static llvm::cl::opt<bool> clUseFastMinMaxOps(
         "Use `arith.minf/maxf` instead of `arith.minimumf/maximumf` ops"),
     llvm::cl::init(false));
 
-// TODO(#10820): Delete the flag. This should be a nop pass to default pipeline
-// while tensor.pad op is lowered to fill + insert_slice before Codegen.
-// However, it causes regressions in terms of compilation time. Skip the passes
-// for now.
-static llvm::cl::opt<bool> clEnablePadConsumerFusion(
-    "iree-llvmcpu-enable-pad-consumer-fusion",
-    llvm::cl::desc("Flag to enable the fusion for pad + consumer"),
-    llvm::cl::init(false));
-
 static llvm::cl::opt<bool> clEnableReassociateFpReductions(
     "iree-llvmcpu-reassociate-fp-reductions",
     llvm::cl::desc("Enables reassociation for FP reductions"),

compiler/src/iree/compiler/Codegen/LLVMCPU/test/pipeline_pad_conv_tests.mlir

@@ -1,4 +1,4 @@
-// RUN: iree-opt --pass-pipeline='builtin.module(iree-llvmcpu-select-lowering-strategy, func.func(iree-llvmcpu-lower-executable-target))' --iree-llvmcpu-enable-pad-consumer-fusion --split-input-file %s | FileCheck %s
+// RUN: iree-opt --pass-pipeline='builtin.module(iree-llvmcpu-select-lowering-strategy, func.func(iree-llvmcpu-lower-executable-target))' --split-input-file %s | FileCheck %s
 
 #executable_target_embedded_elf_x86_64_ = #hal.executable.target<"llvm-cpu", "embedded-elf-x86_64", {cpu = "generic", cpu_features = "+avx512f", data_layout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128", native_vector_size = 32 : index, target_triple = "x86_64-none-elf"}>
 module {