refactor: change tvm wrapper symbol names (#141)

This PR unifies the naming convention of tvm wrapper symbols
@MasterJH5574 .

src/tvm_wrapper.cu

@@ -729,23 +729,22 @@ void _FlashInferBatchQKApplyRotaryInPlace(DLTensor* q, DLTensor* k, DLTensor* in
       })});
 }
 
-// TODO(Zihao): Unify the symbol names
-TVM_REGISTER_GLOBAL("paged_kv_cache.attention_kernel_prefill")
+TVM_REGISTER_GLOBAL("flashinfer.attention_kernel_prefill_with_paged_kv_cache")
     .set_body_typed(_FlashInferAttentionPrefillWithPagedKVCache);
 
-TVM_REGISTER_GLOBAL("paged_kv_cache.attention_kernel_prefill_begin_forward")
+TVM_REGISTER_GLOBAL("flashinfer.attention_kernel_prefill_with_paged_kv_cache_begin_forward")
     .set_body_typed(_FlashInferAttentionPrefillWithPagedKVCacheBeginForward);
 
-TVM_REGISTER_GLOBAL("paged_kv_cache.attention_kernel_prefill_end_forward")
+TVM_REGISTER_GLOBAL("flashinfer.attention_kernel_prefill_with_paged_kv_cache_end_forward")
     .set_body_typed(_FlashInferAttentionPrefillWithPagedKVCacheEndForward);
 
-TVM_REGISTER_GLOBAL("paged_kv_cache.attention_kernel_decode")
+TVM_REGISTER_GLOBAL("flashinfer.attention_kernel_decode_with_paged_kv_cache")
     .set_body_typed(_FlashInferAttentionDecodeWithPagedKVCache);
 
-TVM_REGISTER_GLOBAL("paged_kv_cache.attention_kernel_decode_begin_forward")
+TVM_REGISTER_GLOBAL("flashinfer.attention_kernel_decode_with_paged_kv_cache_begin_forward")
     .set_body_typed(_FlashInferAttentionDecodeWithPagedKVCacheBeginForward);
 
-TVM_REGISTER_GLOBAL("paged_kv_cache.attention_kernel_decode_end_forward")
+TVM_REGISTER_GLOBAL("flashinfer.attention_kernel_decode_with_paged_kv_cache_end_forward")
     .set_body_typed(_FlashInferAttentionDecodeWithPagedKVCacheEndForward);
 
 TVM_REGISTER_GLOBAL("flashinfer.attention_kernel_prefill_with_ragged_kv_cache")