Enable KVCache reuse for disagg

Signed-off-by: Iman Tabrizian <10105175+tabrizian@users.noreply.github.com>

Author: Tabrizian

tests/integration/defs/accuracy/test_disaggregated_serving.py

@@ -960,7 +960,6 @@ def test_auto_dtype(self, overlap_scheduler):
         ctx_server_config["cache_transceiver_config"] = {"backend": "DEFAULT"}
         gen_server_config["cache_transceiver_config"] = {"backend": "DEFAULT"}
         ctx_server_config["kv_cache_config"] = {
-            "enable_block_reuse": False,
             "free_gpu_memory_fraction": 0.7,
             "tokens_per_block": 64,
             "dtype": "fp8"
@@ -977,7 +976,6 @@ def test_auto_dtype(self, overlap_scheduler):
         ctx_server_config["enable_attention_dp"] = True
         ctx_server_config["enable_autotuner"] = False
         gen_server_config["kv_cache_config"] = {
-            "enable_block_reuse": False,
             "tokens_per_block": 64,
             "free_gpu_memory_fraction": 0.7,
             "dtype": "fp8"