feat: add embedding types, port and openai-compat transport

src/polygateway/types.py

@@ -186,3 +186,30 @@ class GlobalLimits:
     def __post_init__(self) -> None:
         if self.max_concurrency < 0 or self.rpm < 0 or self.tpm < 0:
             raise ValueError("全局限额不能为负(0 表示不启用)")
+
+
+@dataclass(frozen=True)
+class EmbeddingTransportResult:
+    """一次原始 embedding 调用的解析结果(M2 设计 §7.2;transport → client)。"""
+
+    vectors: list[list[float]]
+    dim: int
+    prompt_tokens: int
+    usage_source: str  # measured | estimated
+    raw: dict[str, Any]
+
+
+@dataclass(frozen=True)
+class EmbeddingResponse:
+    """一次治理 embedding 调用的统一响应(多批合并;与输入等长保序)。"""
+
+    vectors: list[list[float]]
+    dim: int
+    model: str
+    provider: str
+    prompt_tokens: int
+    usage_source: str
+    latency_ms: int
+    call_id: str
+    source_name: str
+    cost: float | None = None