fix(pytorch): pass --disable-ngc to EFA installer for 1.48+

Starting with EFA 1.48, the installer auto-detects NGC containers via
/opt/nvidia/nvidia_entrypoint.sh (present in nvidia/cuda:*-amzn2023 base
images) and skips installing the AL2023 libnccl-ofi RPM. This left the
pytorch runtime image without libnccl-net.so, breaking every unit /
sanity / single-gpu / efa test that asserts the OFI plugin is present.

The sibling scripts/docker/common/install_efa_amzn2023.sh already has
this fix (lines 37-48); mirror the version-gated --disable-ngc flag here
so aws-ofi-nccl installs regardless of the base-image NGC detection.

Only affects pytorch - sglang, vllm, and vllm-omni use the common
installer via their own paths.

Author: Bhanu Teja Goshikonda

scripts/docker/pytorch/install_efa.sh

@@ -10,7 +10,13 @@ mkdir -p /tmp/efa && cd /tmp/efa
 curl -O "https://s3-us-west-2.amazonaws.com/aws-efa-installer/aws-efa-installer-${EFA_VERSION}.tar.gz"
 tar -xf "aws-efa-installer-${EFA_VERSION}.tar.gz"
 cd aws-efa-installer
-./efa_installer.sh -y --skip-kmod --skip-limit-conf --no-verify
+# EFA 1.48+ auto-detects NGC containers (via /opt/nvidia/nvidia_entrypoint.sh
+# in nvidia/cuda:*-amzn2023 bases) and skips the AL2023 libnccl-ofi RPM.
+# Force the standard install with --disable-ngc so aws-ofi-nccl gets installed.
+ver_ge() { [ "$1" = "$2" ] || [ "$2" = "$(printf '%s\n' "$1" "$2" | sort -V | head -n1)" ]; }
+EFA_EXTRA_ARGS=""
+ver_ge "$EFA_VERSION" "1.48.0" && EFA_EXTRA_ARGS="--disable-ngc"
+./efa_installer.sh -y --skip-kmod --skip-limit-conf --no-verify ${EFA_EXTRA_ARGS}
 rm -rf /tmp/efa
 
 # Configure OpenMPI — allow root execution