fix(ci): update est_time for 57 tests based on runtime analysis (#21896)

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

test/registered/moe/test_cutedsl_moe.py

@@ -12,7 +12,7 @@ from sglang.srt.layers.moe.flashinfer_cutedsl_moe import flashinfer_cutedsl_moe_
 from sglang.srt.layers.moe.topk import TopKConfig, select_experts
 from sglang.test.ci.ci_register import register_cuda_ci
 
-register_cuda_ci(est_time=300, suite="stage-c-test-4-gpu-b200")
+register_cuda_ci(est_time=20, suite="stage-c-test-4-gpu-b200")
 
 SKIP_TEST = torch.cuda.get_device_capability() < (10, 0)
 SKIP_REASON = "Nvfp4 Requires compute capability of 10 or above."

test/registered/moe/test_moe_ep.py

@@ -12,7 +12,7 @@ from sglang.test.test_utils import (
     popen_launch_server,
 )
 
-register_cuda_ci(est_time=140, suite="stage-b-test-2-gpu-large")
+register_cuda_ci(est_time=250, suite="stage-b-test-2-gpu-large")
 
 
 class TestEp(CustomTestCase):

test/registered/moe/test_triton_fused_moe.py

@@ -12,7 +12,7 @@ from sglang.srt.server_args import ServerArgs, set_global_server_args_for_schedu
 from sglang.test.ci.ci_register import register_cuda_ci
 from sglang.test.test_utils import CustomTestCase
 
-register_cuda_ci(est_time=89, suite="stage-b-test-1-gpu-large")
+register_cuda_ci(est_time=30, suite="stage-b-test-1-gpu-large")
 
 
 class TestFusedMOE(CustomTestCase):