[Test] Consolidate test cleanup and CI taxonomy (net -11.4K lines) (#37436)

Co-authored-by: Mick Qian <mickqian@users.noreply.github.com>

scripts/ci/utils/diffusion/diffusion_case_parser.py

@@ -44,7 +44,9 @@ STARTUP_OVERHEAD_SECONDS = 120.0
 # Paths relative to repository root
 BASELINE_REL_PATH = "python/sglang/multimodal_gen/test/server/perf_baselines"
 BASELINE_PLATFORM_ORDER = ("h100", "b200", "5090")
-RUN_SUITE_REL_PATH = "python/sglang/multimodal_gen/test/run_suite.py"
+RUN_SUITE_REL_PATH = (
+    "python/sglang/multimodal_gen/test/runner/diffusion_suite_runner.py"
+)
 
 USE_NPU_CONFIGS = os.getenv("USE_NPU_CONFIGS", "0").lower() in ("1", "true")
 

scripts/ci/utils/diffusion/verify_diffusion_coverage.py

@@ -155,7 +155,8 @@ def print_missing_standalone_estimates_summary(
 
     print("\n" + "=" * 60)
     print(
-        "Add standalone estimate(s) to python/sglang/multimodal_gen/test/run_suite.py"
+        "Add standalone estimate(s) to "
+        "python/sglang/multimodal_gen/test/runner/diffusion_suite_runner.py"
     )
     print("=" * 60)
     print("The following standalone file(s) used fallback estimate 300.0s.")