[Speculative Decoding] Unify Spec and non-spec branch (#6685)

* optimize spec-inference architecture

* delete debug log

* optimize spec_method usage  && fix unit_test

* add claude unit-test skill

* fix some ugly bug

* enhance robustness and bounds check

* unify method & spec_method to method to avoid bug

* activate CI

* fix unit test

* Unify logprobs computation for naive and speculative decoding, fix CUDA kernel

* fix logprob bug && optimize verify kernel

* fix exist_decode() judge

fastdeploy/spec_decode/suffix.py

@@ -14,13 +14,17 @@
 # limitations under the License.
 """
 
+from typing import TYPE_CHECKING
+
 import numpy as np
 
-from fastdeploy.config import FDConfig
 from fastdeploy.utils import spec_logger
 
 from .base import Proposer
 
+if TYPE_CHECKING:
+    from fastdeploy.config import FDConfig
+
 try:
     from arctic_inference.suffix_decoding import SuffixDecodingCache
 except ImportError:
@@ -34,7 +38,7 @@ class SuffixProposer(Proposer):
     Uses SuffixDecodingCache to generate draft tokens based on suffix tree matching.
     """
 
-    def __init__(self, fd_config: FDConfig):
+    def __init__(self, fd_config: "FDConfig"):
         super().__init__(fd_config)
 
         if SuffixDecodingCache is None: