Fixed the issue in which GPUDirect-TCPX for a3-highgpu-8g machine types was

## Fix Fixed Fixed the issue in which GPUDirect-TCPX for `a3-highgpu-8g` machine types was incompatible with the Linux kernel version that was used by Container-Optimized OS in GKE version 1.34 and later. To prevent errors, GKE blocked creating or upgrading node pools that used the `a3-highgpu-8g` machine type to version 1.34 or later. For more information about this issue, see [GKE known issues](https://cloud.google.com/kubernetes-engine/docs/troubleshooting/known-issues#tcpx-cos125). You can now create or upgrade node pools that use the `a3-highgpu-8g` machine type to any of the following GKE versions. **Automatic upgrades of these node pools from version 1.33 to version 1.34 or later are no longer blocked.** * For minor version 1.34, use patch version 1.34.5-gke.1153000 or later. * For minor version 1.35, use patch version 1.35.2-gke.1485000 or later. * For minor version 1.36 and later, use any available patch version. In GKE version 1.34 and later, you must use version 3.1.9 or later of the GPUDirect-TCPX installer and version 2.0.12 or later of the GPUDirect-TCPX sidecar. If you previously installed these components, verify that the container images use these versions or later. **To avoid degraded performance or workload failures, update your installer and sidecar image versions before the`a3-highgpu-8g` node pools are manually or automatically upgraded to version 1.34 or later.** These container image versions correspond to the upstream definitions maintained in the [gpudirect-tcpx GitHub repository](https://github.com/GoogleCloudPlatform/container-engine-accelerators/tree/master/gpudirect-tcpx).