refactor(e2e): remove Jetson dispatch backend

.agents/skills/nemoclaw-maintainer-e2e/SKILL.md

@@ -67,7 +67,7 @@ Choose exactly one mode:
   - every default-selected free-standing workflow E2E except `Exact staging Brev Launchable`;
   - every shared credential-free test; and
   - these controller-selected registry targets: `ubuntu-policy-custom-missing-presets-negative`, `ubuntu-repo-cloud-langchain-deepagents-code`, `ubuntu-repo-cloud-openclaw`, and `ubuntu-repo-docker-post-reboot-recovery`.
-  The run skips `jetson-nvmap-gpu` unless its Colossus dispatch flag is `true`.
+  The run skips `jetson-nvmap-gpu` unless `allow_jetson_dispatch` is `true`.
   It skips `llama-cpp-dgx-spark-plan` and `llama-cpp-dgx-spark-qualification` unless their runner-queue flag is `true`.
 - For protected managed-image runtime qualification, set `E2E_JOBS=managed-image-protected-runtime`. The exact candidate must contain `ci/protected-managed-image-multiarch-activation-v1.json` and `ci/protected-managed-image-runtime-activation-v1.json`.
 
@@ -243,12 +243,11 @@ repository `maintain` or `admin` permission before the Launchable path's source
 checkout. That role check is the authorization.
 A user permitted to dispatch this workflow may set `allow_jetson_dispatch=true`
 to add `jetson-nvmap-gpu` to an empty-selector manual run or enable its explicit
-selection. Set it only after every deployment check in
-[Jetson Dispatch Through Colossus](../../../test/e2e/docs/jetson-colossus-dispatch.md)
-passes.
-The Jetson lifecycle removes and verifies only its fixed job-owned cleanup
-allowlist. Do not treat `cleanup: "succeeded"` as evidence that every possible
-candidate host change was reversed.
+selection. Set it only after the operator-owned service is available and
+compatible with HTTP contract
+version `1.0.0`, and `JETSON_DISPATCH_URL` contains its verified HTTPS origin.
+See [Jetson Dispatch Controller](../../../test/e2e/docs/jetson-dispatch.md).
+Require the uploaded Jetson receipt to report `cleanup: "succeeded"`.
 A permitted dispatcher may set `allow_dgx_spark_runner_queue=true` to add
 `llama-cpp-dgx-spark-plan` and `llama-cpp-dgx-spark-qualification` to an
 empty-selector manual run or enable explicit qualification selection. Set it

.github/workflows/e2e.yaml

@@ -34,7 +34,7 @@ on:
           - internal-nvidia
           - public-nvidia
       allow_jetson_dispatch:
-        description: "Set true for jetson-nvmap-gpu only after the Colossus dispatcher, Cloudflare Tunnel, cleanup helper, and JETSON_DISPATCH_URL repository variable have passed the deployment checks in test/e2e/docs/jetson-colossus-dispatch.md."
+        description: "Set true for jetson-nvmap-gpu only after the operator-owned dispatch backend is available and JETSON_DISPATCH_URL contains its verified HTTPS origin. Refer to test/e2e/docs/jetson-dispatch.md."
         required: false
         default: false
         type: boolean
@@ -4924,7 +4924,7 @@ jobs:
     needs: generate-matrix
     if: ${{ inputs.allow_jetson_dispatch && github.repository == 'NVIDIA/NemoClaw' && github.ref == 'refs/heads/main' && github.event_name == 'workflow_dispatch' && (inputs.checkout_repository == '' || inputs.checkout_repository == github.repository) && ((inputs.jobs == '' && inputs.targets == '') || contains(format(',{0},', inputs.jobs), ',jetson-nvmap-gpu,') || contains(format(',{0},', inputs.targets), ',jetson-nvmap-gpu,')) }}
     concurrency:
-      group: jetson-nvmap-gpu-colossus
+      group: jetson-nvmap-gpu-dispatch
       queue: max
       cancel-in-progress: false
     runs-on: ubuntu-latest
@@ -4945,7 +4945,7 @@ jobs:
         with:
           node-version: 22
 
-      - name: Dispatch exact commit to Jetson through Colossus
+      - name: Dispatch exact commit to Jetson through operator backend
         env:
           E2E_ARTIFACT_DIR: ${{ runner.temp }}/e2e-artifacts/live/jetson-nvmap-gpu
           JETSON_DISPATCH_CANDIDATE_SHA: ${{ inputs.checkout_sha || github.sha }}

test/e2e/README.md

@@ -456,12 +456,11 @@ runs remain queued instead of replacing one another.
 
 The Jetson nvmap and DGX Spark llama.cpp jobs remain excluded from ordinary and
 full runs unless their independent opt-in flags are `true`.
-Set `allow_jetson_dispatch=true` to select `jetson-nvmap-gpu` only after every
-deployment check in
-[Jetson Dispatch Through Colossus](docs/jetson-colossus-dispatch.md) passes.
-The Colossus lifecycle removes and verifies only the fixed job-owned cleanup
-allowlist. Its cleanup evidence does not attest that every possible candidate
-host change was reversed.
+Set `allow_jetson_dispatch=true` to select `jetson-nvmap-gpu` after the
+operator-owned dispatch service is available at the repository variable
+`JETSON_DISPATCH_URL`. Refer to the
+[Jetson dispatch controller](docs/jetson-dispatch.md) for the trusted workflow,
+HTTP contract, and evidence boundary that NemoClaw owns.
 Set `allow_dgx_spark_runner_queue=true` to select both
 `llama-cpp-dgx-spark-plan` and `llama-cpp-dgx-spark-qualification`.
 GitHub can pause the qualification job for the
@@ -798,7 +797,7 @@ For a PR revision run, a repository maintainer or administrator leaves `jobs` an
 - every shared credential-free test; and
 - these controller-selected registry targets: `ubuntu-policy-custom-missing-presets-negative`, `ubuntu-repo-cloud-langchain-deepagents-code`, `ubuntu-repo-cloud-openclaw`, and `ubuntu-repo-docker-post-reboot-recovery`.
 
-The run skips `jetson-nvmap-gpu` unless its Colossus dispatch flag is `true`.
+The run skips `jetson-nvmap-gpu` unless `allow_jetson_dispatch` is `true`.
 It skips `llama-cpp-dgx-spark-plan` and `llama-cpp-dgx-spark-qualification`
 unless their runner-queue flag is `true`.
 The trusted workflow definition remains on `main` and binds the candidate head to the current PR base SHA.

test/e2e/docs/README.md

@@ -314,10 +314,10 @@ test/e2e/
   `jobs` value is `managed-image-protected-runtime`; `targets` must remain empty.
   Refer to [NemoClaw E2E CI](../README.md).
 
-- [Jetson Dispatch Through Colossus](jetson-colossus-dispatch.md) defines how
-  maintainers deploy and verify the temporary authenticated controller,
-  dispatcher, SSH, bounded cleanup, and stale-lock recovery path for
-  `jetson-nvmap-gpu`.
+- [Jetson dispatch controller](jetson-dispatch.md) defines the NemoClaw-owned
+  HTTP contract, trusted GitHub controller, repository configuration, and
+  evidence for `jetson-nvmap-gpu`. The service behind that contract is
+  operator-owned infrastructure.
 
 - `.github/workflows/e2e.yaml` runs selected or all supported
   live E2E targets and uploads an explicit artifact allowlist with