[sivanantha321]

Description

This adds the OpenAI-compatible Batch API (/v1/batches) on top of the Files API, with the same
single-backend design. Clients can create, retrieve, list, and cancel batches through the gateway,
and download batch results through the existing Files API.

Like a file, a batch lives on one upstream backend: the backend that holds its input file. The
gateway keeps no state. It encrypts the owning backend and the routing model into each batch id it
returns, using the same codec and fileIDEncryptionSeed as file ids, so no new configuration is
needed.

Routing:

• Create (POST /v1/batches) takes no model. It is routed by the backend and model recorded in
  the gateway input_file_id, which is replaced by the provider's own file id before forwarding.
• Retrieve (GET /v1/batches/{id}) and cancel (POST /v1/batches/{id}/cancel) are routed by the
  model recorded in the batch id, and the path is rewritten to the provider's id.
• List (GET /v1/batches) is routed by ?model=, or by the model in a gateway after cursor, so
  SDK auto-pagination works after the first page. model is not forwarded upstream.

Responses:

• id, first_id, and last_id become gateway batch ids.
• input_file_id, output_file_id, and error_file_id become gateway file ids for the same
  backend and model, so GET /v1/files/{output_file_id}/content works unchanged.
• Ids are rewritten in place, so fields the gateway does not model (for example usage) pass
  through. Upstream error responses pass through unchanged. A successful response whose ids cannot
  be re-encoded becomes a 502 rather than leaking provider ids.

As with Files, each route rule must have exactly one backend, the id must belong to the backend
serving the request, and only the OpenAI schema has translators today.

Every request goes through the Batch translator in the upstream filter, as the Files processor
does, so a future provider translator can rewrite requests.

Related PRs

• Depends on feat: support for OpenAI Files API with single backend #2884

Special notes for reviewers (if applicable)

• Out of scope in this PR: multi-backend routing, non-OpenAI translators, rewriting model inside the JSONL
  input file, and token-usage metering. Batch work runs asynchronously on the provider, so the
  gateway does not record token usage for it.

[netlify]

✅ Deploy Preview for theagentrouter ready!

Name	Link
🔨 Latest commit	194dfbd
🔍 Latest deploy log	https://app.netlify.com/projects/theagentrouter/deploys/6aeef6b2665214227bad5bf9
😎 Deploy Preview	https://deploy-preview-2883--theagentrouter.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify project configuration.

[codecov]

Codecov Report

❌ Patch coverage is 94.13408% with 21 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
internal/extproc/batches_processor.go	93.09%	19 Missing ⚠️
internal/extproc/idrouting_helpers.go	96.92%	2 Missing ⚠️

📢 Thoughts on this report? Let us know!