chore: require python 3.12 and adopt PEP 695 type parameters

The telemetry write budget needs asyncio.timeout, whose uncancel accounting
was only fixed after 3.11.1 — pinning the floor at 3.12 removes that hazard
instead of working around it.

Raising ruff's target-version turns on UP047, so gather_bounded,
_anext_within and stream_with_liveness_timeouts move to def f[T](...) and
the two module-level TypeVars go away. That syntax is a SyntaxError on
3.11, so it can only land together with the version bump.

src/polygateway/client.py

@@ -13,7 +13,7 @@ import hashlib
 import json
 import random
 import time
-from typing import TYPE_CHECKING, Any, Literal, TypeVar
+from typing import TYPE_CHECKING, Any, Literal
 
 from polygateway.backends.memory.breaker import InMemoryGate
 from polygateway.backends.memory.cache import InMemoryCache
@@ -63,8 +63,6 @@ if TYPE_CHECKING:
         SourceConfig,
     )
 
-_T = TypeVar("_T")
-
 
 def _guard_thinking(
     sources: list[SourceConfig],
@@ -442,7 +440,7 @@ def _build_structured(
         return None, None
 
 
-async def gather_bounded(aws: Iterable[Awaitable[_T]], *, concurrency: int) -> list[_T]:
+async def gather_bounded[T](aws: Iterable[Awaitable[T]], *, concurrency: int) -> list[T]:
     """有界并发 gather(D5 便利函数,替代 VT 手搓 semaphore+gather 样板)。
 
     语义与 `asyncio.gather` 默认一致: 结果保序、首个异常上抛;仅增加并发上限。
@@ -451,7 +449,7 @@ async def gather_bounded(aws: Iterable[Awaitable[_T]], *, concurrency: int) -> l
         raise ValueError("concurrency 必须 ≥ 1")
     sem = asyncio.Semaphore(concurrency)
 
-    async def _run(aw: Awaitable[_T]) -> _T:
+    async def _run(aw: Awaitable[T]) -> T:
         async with sem:
             return await aw
 

src/polygateway/streaming.py

@@ -14,13 +14,11 @@ from __future__ import annotations
 import asyncio
 import contextlib
 import time
-from typing import TYPE_CHECKING, TypeVar
+from typing import TYPE_CHECKING
 
 if TYPE_CHECKING:
     from collections.abc import AsyncIterator
 
-_T = TypeVar("_T")
-
 
 class StreamLivenessTimeout(Exception):  # noqa: N818 — 三项目冻结的公共名
     """流活性超时异常。
@@ -38,14 +36,14 @@ class StreamLivenessTimeout(Exception):  # noqa: N818 — 三项目冻结的公
         super().__init__(f"流活性超时({kind}, elapsed={elapsed_s:.1f}s)")
 
 
-async def _anext_within(
-    it: AsyncIterator[_T],
+async def _anext_within[T](
+    it: AsyncIterator[T],
     timeout_s: float,
     *,
     kind: str,
     start: float,
     first: bool,
-) -> _T:
+) -> T:
     """限时取下一项;本层 deadline 触发抛 StreamLivenessTimeout(kind)。
 
     上游自抛的 TimeoutError 用 cm.expired() 区分,原样上抛不误吞。
@@ -59,13 +57,13 @@ async def _anext_within(
         raise StreamLivenessTimeout(kind, time.monotonic() - start, not first) from None
 
 
-async def stream_with_liveness_timeouts(
-    source: AsyncIterator[_T],
+async def stream_with_liveness_timeouts[T](
+    source: AsyncIterator[T],
     *,
     ttft_s: float,
     inter_token_s: float,
     total_s: float,
-) -> AsyncIterator[_T]:
+) -> AsyncIterator[T]:
     """逐项产出 source,并施加三层活性超时。
 
     关键实现: 超时**只包裹单次 __anext__**,绝不包裹 yield——否则总时长