Make PredictedLatencyScorer PD Aware

[RishabhSaini]

1. Added configurable EndpointRoleLabel parameter to EPP config to enable role-aware predictions for any deployment (e.g., "llm-d.ai/role" for prefill/decode in llm-d).

 - type: predicted-latency-scorer
         parameters:
           endpointRoleLabel: "llm-d.ai/role"
           sloBufferFactor: 1.0
           headroomSelectionStrategy: "least"

3. Created helper functions buildPredictionRequest() and buildTrainingEntry() that conditionally populate PodType from endpoint labels for role-specific model training (prefill vs decode).

4. Changed runningRequestLists from map to sync.Map for thread safety. Updated all access patterns.

5. Implemented prefill pod tracking across PreRequest (runningRequest tracking), ResponseReceived (training data collection), and ResponseComplete (runningRequest cleanup) hooks. Prefill TTFT is calculated as ResponseReceived - RequestReceived timestamp, capturing scheduling, queuing, prefill KV cache processing time, and network hop for KV cache transfer to decode pod.

The corresponding PR in llm-d-inference-scheduler is no longer needed:
Refer to llm-d/llm-d-inference-scheduler#564 (comment)

[netlify]

✅ Deploy Preview for gateway-api-inference-extension ready!

Name	Link
🔨 Latest commit	aa4549a
🔍 Latest deploy log	https://app.netlify.com/projects/gateway-api-inference-extension/deploys/698fad8e93791b0008d48906
😎 Deploy Preview	https://deploy-preview-2145--gateway-api-inference-extension.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.

[k8s-ci-robot]

Hi @RishabhSaini. Thanks for your PR.

I'm waiting for a kubernetes-sigs member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

Details

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.

[ahg-g]

/ok-to-test

I am concerned that we are still not cleaning up all the naming and terminology around the predicted-latency work (see #2032), some of it is not aligned with the terminology we have adopted so far (like using the term "router" or "slo"), and it is causing confusion in discussions with the community.

[kaushikmitr]

/ok-to-test

I am concerned that we are still not cleaning up all the naming and terminology around the predicted-latency work (see #2032), some of it is not aligned with the terminology we have adopted so far (like using the term "router" or "slo"), and it is causing confusion in discussions with the community.

the naming of the router is resolved now

[kaushikmitr]

/retest

[kaushikmitr]

@RishabhSaini lgtm, one question, we recently updated the latencypredictor plugin to move the prediction step to preparedata step, which happens before scheduling and introduced admission plugin. Just conforming your change still works with those changes. https://github.com/kubernetes-sigs/gateway-api-inference-extension/pulls?q=is%3Apr+is%3Aclosed+kaushikmitr