Move the panorama keypoint detector onto the engine, and probe with a detector

XFeat's two exports are a Keypoints role now; the crate no longer names
tract, and the app compiles TensorRT engines for both ahead of the
first merge. The probe picks the smallest *detector* rather than the
smallest file: the tablet's first run chose the 112 KB eye classifier,
which has no int8 form, and reported the Hexagon as failed for want of
one.

ui/dr-ui/src/inference.rs

@@ -38,7 +38,14 @@ pub fn init(runtime_dirs: Vec<PathBuf>) {
         runtime_dirs,
         cache_dir: crate::library::inference_cache_dir(),
         models,
-        embedded: vec![(Role::Segmenter, dr_segment::embedded_model_bytes())],
+        embedded: {
+            let [landscape, portrait] = dr_pano::xfeat::embedded_model_bytes();
+            vec![
+                (Role::Segmenter, dr_segment::embedded_model_bytes()),
+                (Role::Keypoints, landscape),
+                (Role::Keypoints, portrait),
+            ]
+        },
         ceiling: None,
         threads: 0,
         decay: std::time::Duration::ZERO,