feat: add gateway client with env-driven assembly

Includes config aggregation for multi-source env keys, from_env and
from_settings factories with explicit shared-backend injection,
gather_bounded, top-level exports, tightened import-linter layers with
the gate removed from the Makefile, and the finalized .env.example.

tests/contracts/conftest.py

@@ -25,10 +25,14 @@ class FakeClock:
 
 
 def make_source(name: str = "s1", **overrides) -> SourceConfig:
-    base = dict(
-        name=name, provider="openai", base_url="https://gw.example/v1",
-        api_key="sk-test", model="m", timeout_s=10.0,
-    )
+    base = {
+        "name": name,
+        "provider": "openai",
+        "base_url": "https://gw.example/v1",
+        "api_key": "sk-test",
+        "model": "m",
+        "timeout_s": 10.0,
+    }
     base.update(overrides)
     return SourceConfig(**base)
 
@@ -44,8 +48,11 @@ def limiter_factory(request, clock):
 
     def make(sources: list[SourceConfig], global_limits: GlobalLimits, lease_ttl_s: float = 100.0):
         return InMemoryLimiter(
-            scope="llm", sources={s.name: s for s in sources},
-            global_limits=global_limits, lease_ttl_s=lease_ttl_s, now=clock,
+            scope="llm",
+            sources={s.name: s for s in sources},
+            global_limits=global_limits,
+            lease_ttl_s=lease_ttl_s,
+            now=clock,
         )
 
     return make