refactor: remove inference redundancy and fix cache leaks

- drop Executor unused tokenizer field, _head_dim, stale metrics docstring
- unify greedy sampling via SamplingPipeline.sample, drop top-level duplicate
- drop Task.flush_remaining no-op and unreachable prompt-length branch
- drop ProtocolHandler redundant chunks list (reuse body)
- fix page_size=1 token-slot leak on task_free
- clear _task_pages/_task_slots on alloc-failure paths
- reset _bind_state on task_free to avoid stale steady-state reuse
- remove unreachable contiguous branches in paged-only helpers

astrai/inference/core/task.py

@@ -1,4 +1,3 @@
-import logging
 import threading
 import time
 import uuid
@@ -11,8 +10,6 @@ from tokenizers.decoders import DecodeStream
 from astrai.inference.core.metrics import MetricsCollector
 from astrai.tokenize.tokenizer import AutoTokenizer
 
-logger = logging.getLogger(__name__)
-
 STOP = object()
 
 
@@ -92,16 +89,6 @@ class Task:
             self._decoder = StreamDecoder(tokenizer)
         return self._decoder.push(self.output_ids[-1])
 
-    def flush_remaining(self, tokenizer: AutoTokenizer) -> str:
-        """Emit any text still buffered in the decoder.
-
-        With the Rust-native DecodeStream, the stream is always in a
-        correct state — any completed text was already emitted by the
-        last ``push``.  A trailing incomplete multi-byte sequence has no
-        valid text to emit, so this is a no-op.
-        """
-        return ""
-
     @property
     def next_pos(self) -> int:
         # The first output is sampled from prefill and enters KV on the next step.
@@ -157,11 +144,6 @@ class TaskManager:
         if len(prompt_ids) > self.max_seq_len:
             prompt_ids = prompt_ids[-self.max_seq_len :]
 
-        if len(prompt_ids) > self.max_seq_len:
-            if stream_callback:
-                stream_callback(STOP)
-            return task_id
-
         if max_tokens is None:
             max_tokens = self.max_seq_len - len(prompt_ids)
         else: