Fix sagemaker-entrypoint* & remove SageMaker and Vertex from Dockerfile*

[alvarobartt]

What does this PR do?

This PR fixes the AWS SageMaker entrypoints in sagemaker-entrypoint.sh and sagemaker-entrypoint-cuda-all.sh, as those were missing the exec when running the text-embeddings-router process meaning it was being assigned a PID other than 1, so that the signals as the graceful handling (Ctrl + C) were not being propagated into the process, hence not stopping the process.

Besides that, this PR removes the AWS SageMaker stages from both Dockerfile (CPU) and Dockerfile-cuda-all (NVIDIA GPU) as it's not used, since those images are ported as-is into https://github.com/awslabs/llm-hosting-container/tree/main/huggingface/pytorch/tei/docker and re-built from scratch, so there's not a clear benefit of including the stage on the build here.

Finally, this PR also removes the BUILD_ARG for Google Cloud Vertex AI, as well as the conditional if-else when building the text-embeddings-router in Dockerfile-cuda-all that was building the binary with the --features google, which is not required as the Dockerfile is also ported as-is into https://github.com/huggingface/Google-Cloud-Containers/tree/main/containers/tei, meaning the BUILD_ARG is not being used here.

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline?
• Was this discussed/approved via a GitHub issue or the forum? Please add a link to it if that's the case.
• Did you make sure to update the documentation with your changes? Here are the documentation guidelines.
• Did you write any new necessary tests? If applicable, did you include or update the insta snapshots?

Who can review?

• @Narsil to ensure the aforementioned changes are fair and make sense
• @arjkesh and @pagezyhf (feel free to ping anyone else for confirmation if needed) to verify that there is no inconvenient on removing the stage for AWS SageMaker in the Dockerfile and Dockerfile-cuda-all here, as those are ported and rebuilt for AWS anyway

Finally, as per the Google Cloud stuff, I can verify from a technical standpoint as per the containers released on Google Cloud that this is not an issue, since those are ported and rebuilt elsewhere.