fix: lower gpu-memory-utilization to 0.85 — prevent profiling OOM

CCCL CachingDeviceAllocator reserves some memory for its bin cache.
With 0.90 utilization + max-model-len=80000, profiling stage OOMs.
0.85 leaves ~1.6GB headroom per GPU for profiling + allocator cache.

computility-run.yaml

@@ -10,7 +10,7 @@ command:
     - --max-model-len
     - '80000'
     - --gpu-memory-utilization
-    - '0.90'
+    - '0.85'
     - --trust-remote-code
     - -tp
     - '4'