[Typo] Remove disable_cache in some tests (tile-ai#755)

* Update test parameters and remove debug print statement

- Adjusted test cases in `test_tilelang_dynamic_symbolic_bench.py` to use smaller matrix sizes (1024x1024) for improved performance and quicker execution.
- Removed a debug print statement from `phase.py` to clean up the code and enhance clarity.

* Refactor loop stack management in warp_specialized_rewriter

- Introduced a new `LoopInfo` struct to encapsulate loop variable details, including `loop_var`, `extent`, and `min`, enhancing clarity and maintainability.
- Updated the `loop_stack_` to utilize `LoopInfo` instead of a pair, improving type safety and readability.
- Adjusted linear index calculations to account for the new structure, ensuring correct behavior in loop transformations.

* Remove unused `torch.backends` import and `tilelang.disable_cache()` calls from multiple test files to enhance code clarity and maintainability.

Author: LeiWang1999

testing/python/cpu/test_tilelang_cpu_gemm.py

@@ -4,8 +4,6 @@
 import tilelang.language as T
 import torch
 
-tilelang.disable_cache()
-
 
 def matmul(M, N, K, block_M, block_N, block_K, dtype="float16", accum_dtype="float"):
     num_stages = 0

testing/python/dynamic/test_tilelang_dynamic_symbolic_bench.py

@@ -1,12 +1,8 @@
 import torch
-import torch.backends
 from tilelang import tvm as tvm
 import tilelang.testing
 import tilelang.language as T
 
-tilelang.testing.set_random_seed(0)
-tilelang.disable_cache()
-
 
 def tl_matmul_block_static(
     M,

testing/python/kernel/test_tilelang_kernel_int4_gemm_mma.py

@@ -1,5 +1,4 @@
 import torch
-import torch.backends
 import tilelang
 from tilelang import tvm as tvm
 import tilelang.testing
@@ -14,7 +13,6 @@
 from tilelang.transform import simplify_prim_func
 
 tilelang.testing.set_random_seed(42)
-tilelang.disable_cache()
 
 
 # @simplify_prim_func

testing/python/language/test_tilelang_language_annotate_pad.py

@@ -3,8 +3,6 @@
 import tilelang.testing
 import torch
 
-tilelang.disable_cache()
-
 
 # add decorator @tilelang.jit if you want to return a torch function
 # @tilelang.jit

testing/python/transform/test_tilelang_transform_config_index_bitwidth.py

@@ -3,8 +3,6 @@
 import tilelang
 import tilelang.language as T
 
-tilelang.disable_cache()
-
 
 def blocksparse_flashattn(batch, heads, seq_len, dim, downsample_len, is_causal):
     block_M = 64

testing/python/transform/test_tilelang_transform_thread_sync.py

@@ -189,4 +189,4 @@ def expected(A: T.Buffer((8192,), "float32")):
 
 
 if __name__ == "__main__":
-    tilelang.disable_cache()
+    tilelang.testing.main()