Add persistent async input scheduler for GEMM kernels (#3520)

Add signal-based synchronization for persistent GEMM kernels where
input data becomes available incrementally. Uses modulo wraparound
(like PyTorch's AsyncMM) for chunk index calculation:
  chunk_idx = ((tile_idx + tile_idx_pivot) / tiles_per_chunk) % num_chunks

Key components:
- PersistentAsyncInputScheduler struct with tiles_per_chunk_m,
  chunk_signals, tile_idx_pivot_m, and num_chunks fields
- wait_eq_wave method using __builtin_amdgcn_s_sleep for power efficiency
- IsSupportedArgument validation for scheduler parameters
- Example demonstrating async input scheduling with simulated producer
- GTest unit tests covering all layout combinations

example/ck_tile/03_gemm/gemm_utils.hpp

@@ -456,7 +456,8 @@ inline auto create_args()
         .insert("json", "0", "0: No Json, 1: Dump Results in Json format")
         .insert("jsonfile", "gemm.json", "json file name to dump results")
         .insert("flush_cache", "true", "flush cache before running the kernel, defaults to true")
-        .insert("rotating_count", "1000", "rotating count, defaults to 1000");
+        .insert("rotating_count", "1000", "rotating count, defaults to 1000")
+        .insert("test_async", "0", "0: normal gemm, 1: test async input scheduler");
     return arg_parser;
 }