[Feature] Add process-local in-memory KV indexer and Router integration (#33370)

Co-authored-by: Wu, Yutong <yutong.wu@amd.com>
Co-authored-by: TianDi101 <ditian12@amd.com>
Co-authored-by: Zhangheng <hzh0425@apache.org>

experimental/sgl-router/tests/proxy/common/cache_aware_fixture.rs

@@ -0,0 +1,39 @@
+// SPDX-FileCopyrightText: Copyright (c) 2026 The SGLang Authors
+// SPDX-License-Identifier: Apache-2.0
+
+//! Shared router config for the cache-aware proxy tests.
+//!
+//! The model id contains `deepseek-v4` so the tokenizer registry auto-attaches the
+//! built-in V4 chat encoder — the engine-equivalent path — with no template fixture.
+
+use sgl_router::config::{
+    ActiveLoadConfig, CacheAwareConfig, Config, DiscoveryBackend, ModelConfig, ObservabilityConfig,
+    PolicyKind, ProxyConfig, ServerConfig, StaticUrlsDiscoveryConfig,
+};
+
+pub const MODEL: &str = "deepseek-v4-tiny";
+
+/// A single-model `cache_aware_zmq` router. Discovery is a placeholder because
+/// every caller installs its own `WorkerRegistry`.
+pub fn config() -> Config {
+    Config {
+        server: ServerConfig {
+            host: "0".into(),
+            port: 0,
+        },
+        observability: ObservabilityConfig::default(),
+        model: ModelConfig {
+            id: MODEL.into(),
+            tokenizer_path: "tests/fixtures/tiny_tokenizer.json".into(),
+            policy: PolicyKind::CacheAwareZmq,
+            circuit_breaker: None,
+            cache_aware: Some(CacheAwareConfig::default()),
+            sticky: None,
+        },
+        discovery: DiscoveryBackend::StaticUrls(StaticUrlsDiscoveryConfig {
+            urls: vec!["http://placeholder:0".into()],
+        }),
+        proxy: ProxyConfig::default(),
+        active_load: ActiveLoadConfig::default(),
+    }
+}

experimental/sgl-router/tests/proxy/common/mod.rs

@@ -3,5 +3,6 @@
 
 //! Shared test harness re-exports.
 
+pub mod cache_aware_fixture;
 pub mod mock_worker;
 pub mod streaming;