[Optimization] Use Shared CachedRequestData Instance Across All Requests (#20232)

Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>

tests/v1/kv_connector/unit/utils.py

@@ -25,7 +25,6 @@ def assert_scheduler_empty(scheduler: Scheduler):
     assert len(scheduler.running) == 0
     assert len(scheduler.finished_req_ids) == 0
     assert len(scheduler.finished_recving_kv_req_ids) == 0
-    assert len(scheduler._cached_reqs_data) == 0
 
     # EncoderCacheManager.
     assert len(scheduler.encoder_cache_manager.freed) == 0