ci: reuse the agent and shallow-clone so the gate is not all overhead

The tests run in 0.62 seconds; the build took 191. Agent scheduling and image
pull were ~54s, a full clone of a five-commit repo ~25s, and pip install
pytest ~55s. None of that is the thing being demonstrated.

A fixed label plus idleMinutes keeps the pod alive between builds. The default
inline template mints a random POD_LABEL per build, so a fresh pod was
guaranteed every time and the pip cache in the workspace volume was always
cold. With reuse, a second run skips scheduling, skips the image pull, and
finds pytest already cached.

The first run after an idle period still pays full price, so a throwaway run
before a demo is worth doing.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

Jenkinsfile

@@ -1,5 +1,26 @@
 // Test gate for the Hermes code-repair demo.
-podTemplate(cloud: 'kubernetes', yaml: """
+//
+// Tuned for a demo, where the wait between pushing a defect and seeing the
+// build go red is watched by people. The tests themselves run in under a
+// second; everything else was overhead:
+//
+//   agent pod scheduling + image pull   ~54s
+//   full clone of a five-commit repo    ~25s
+//   pip install pytest                  ~55s
+//   the actual tests                    ~0.6s
+//
+// A named template with idleMinutes keeps the pod alive between builds, so a
+// second run reuses it: no scheduling, no image pull, and the pip cache in the
+// workspace volume is already warm. The first run after an idle period still
+// pays the full cost, which is why a throwaway run before a demo is worth it.
+podTemplate(
+  cloud: 'kubernetes',
+  // A fixed label is what makes reuse possible at all. The default inline
+  // template mints a random POD_LABEL per build, so every build necessarily
+  // got a fresh pod.
+  label: 'hermes-code-demo-agent',
+  idleMinutes: 60,
+  yaml: """
 apiVersion: v1
 kind: Pod
 spec:
@@ -27,16 +48,28 @@ spec:
           cpu: 100m
           memory: 128Mi
         limits:
-          cpu: 500m
+          cpu: 1000m
           memory: 512Mi
 """) {
-  node(POD_LABEL) {
+  node('hermes-code-demo-agent') {
     container('python') {
       stage('Checkout') {
-        checkout scm
+        // Depth 1 and no tags: the build only ever needs the commit under
+        // test, and fetching five commits of history to run one test file is
+        // time spent on nothing anyone will look at.
+        checkout([
+          $class: 'GitSCM',
+          branches: scm.branches,
+          userRemoteConfigs: scm.userRemoteConfigs,
+          extensions: [
+            [$class: 'CloneOption', shallow: true, depth: 1, noTags: true, timeout: 5],
+          ],
+        ])
       }
       stage('tests') {
-        sh 'pip install --quiet pytest'
+        // Idempotent and near-instant once the pod's pip cache is warm; the
+        // cost is only paid on a cold agent.
+        sh 'pip install --quiet --disable-pip-version-check pytest'
         sh 'python -m pytest -v --junitxml=build/junit.xml'
       }
     }