Add e2e tests for embedding raw flag

[SamMalayek]

🧩 Summary

This PR adds a CI workflow for end-to-end embedding tests.
It marks the first phase of an effort to move an abstraction of the existing examples/llama-embedding logic behind llama-server, so the server can use llama.cpp’s own embedding implementation instead of external (OpenAI) APIs.

🎯 Motivation & Future

llama-server currently supports OpenAI-compatible /embedding requests, but those are not backed by native llama.cpp logic.
This workflow establishes a reproducible test foundation before refactoring the embedding code so that:

• The server can generate embeddings locally.
• --parallel N can support multiple concurrent embedding requests.
• The standalone CLI will remain for lightweight workflows, while the server will use the same shared embedding path for persistent deployments.

⚙️ CI Implementation

• Adds a GitHub Actions job to run embedding E2E tests with cached GGUF models (TinyLlama).
• Verifies embedding output dimensions and deterministic behavior.
• Uses lightweight models for fast CI runs (with an optional large model test).

🧱 Embedding CPP Logic Flow Update

A small cleanup in print_raw_embeddings() improves readability, logic flow, and isolation.
Although minor, this change is modular alongside the CI workflow changes, touching a vertical slice of the embedding flow without altering evaluation, model logic, or any interface. Note that expecting purely small horizontal modularity ossifies software (makes it brittle).

🚀 Next Steps

1. Extend CI test coverage for all embedding endpoints/flags.
2. Abstract core embedding code from examples into a shared utility (e.g. common/embedding_utils.cpp).
3. Integrate that abstraction into llama-server for local "/embedding" requests (while maintaining CLI endpoints and backwards compatibility).
   a. Extend CI coverage for concurrent (--parallel) embedding tests.

(could actually become more than three steps)

Note:
This PR includes a workflow (run-e2e-embedding.yml) for local/fork testing.
It runs only on feature/* branches and is designed to verify embedding e2e tests in forked CI.
Maintainers may integrate or adapt this into the upstream CI configuration if desired.

[ggerganov]

Too much slop.

[SamMalayek]

Too much slop.

This is a pristine PR with a pristine plan. However, I'll interpret your comment as "too much scope creep" after I just landed a PR in a relatively sloppy part of the codebase (examples -- totally fair btw), and I'll push another PR that:

• Removes the embedding.cpp improvement.
• Refactors the embedding tests into a focused, deterministic CLI suite with broader input coverage and reproducible numeric validation.
• Removes e2e test benchmarking (which would have been quite useful for this much-needed refactor of the embedding cli code, but I can just run these locally).

... #16940

Furthermore, I'm opening a discussion RFC as well: #16957. This is a plan that is simply needed for both our Llama.cpp repos (including my cloned repo) because:

• The embedding endpoint for llama-server should be native, rather than use OpenAI's APIs.
• I actually need this for my project's embedding pipeline.