[lbarrosby]

What does this PR do?

Adds a new Agent-based integration for Hugging Face Text Embeddings Inference (TEI).

The check scrapes the Prometheus endpoint TEI serves at /metrics and maps the te_* metrics to hugging_face_tei.*: request counts, successes and failures (the err label is renamed to error_type, as in the TGI integration), end-to-end, tokenization, queue and inference latency histograms, queue size, batching, and the per-input embed (embedding models) and predict (re-ranker and classification models) families.

Motivation

TEI is the embedding and re-ranking backend in many RAG deployments, often running next to TGI or vLLM. The hugging_face_tgi integration in integrations-core covers text generation, but there is no integration for TEI. Metric names follow the same layout as hugging_face_tgi so both can be used side by side.

I'm happy to be listed as the maintainer of this integration.

Review checklist

• PR has a meaningful title
• Feature or bugfix has tests
• Git history is clean
• If PR impacts documentation, docs team has been notified or an issue has been opened on the documentation repo
• If this PR includes a log pipeline, please add a description describing the remappers and processors.

Additional Notes

• Both fixtures are real scrapes from TEI 1.9.4: metrics.txt from an embedding model (sentence-transformers/all-MiniLM-L6-v2) and metrics_rerank.txt from a re-ranker (cross-encoder/ms-marco-MiniLM-L-6-v2). The embed and predict families are exclusive to each model type, so the metadata symmetry test runs the check against both.
• Tested locally with ddev test hugging_face_tei (unit), ddev test --lint, and ddev validate (config, models, metadata, service-checks, readmes, openmetrics, typos). I also ran the check against a live TEI server.
• /metrics is a public route in TEI, so it does not need the API key.