Add RKNN / Object Detection Pipeline (#1144)

Tested on Orange Pi 5 and Cool Pi 4B. Merge with parts of the OpenCV DNN PR. 

Adds support for YOLOv5s models for Rockchip CPUs with a NPU. Right now hard coded to a note model from alex_idk. Very much still incubating and largely untested.

photon-client/src/components/dashboard/tabs/ObjectDetectionTab.vue

@@ -0,0 +1,35 @@
+<script setup lang="ts">
+import { useCameraSettingsStore } from "@/stores/settings/CameraSettingsStore";
+import { PipelineType } from "@/types/PipelineTypes";
+import PvSlider from "@/components/common/pv-slider.vue";
+import { computed, getCurrentInstance } from "vue";
+import { useStateStore } from "@/stores/StateStore";
+
+// TODO fix pipeline typing in order to fix this, the store settings call should be able to infer that only valid pipeline type settings are exposed based on pre-checks for the entire config section
+// Defer reference to store access method
+const currentPipelineSettings = useCameraSettingsStore().currentPipelineSettings;
+
+const interactiveCols = computed(
+  () =>
+    (getCurrentInstance()?.proxy.$vuetify.breakpoint.mdAndDown || false) &&
+    (!useStateStore().sidebarFolded || useCameraSettingsStore().isDriverMode)
+)
+  ? 9
+  : 8;
+</script>
+
+<template>
+  <div v-if="currentPipelineSettings.pipelineType === PipelineType.ObjectDetection">
+    <pv-slider
+      v-model="currentPipelineSettings.confidence"
+      class="pt-2"
+      :slider-cols="interactiveCols"
+      label="Confidence"
+      tooltip="The minimum confidence for a detection to be considered valid. Bigger numbers mean fewer but more probable detections are allowed through."
+      :min="0"
+      :max="1"
+      :step="0.01"
+      @input="(value) => useCameraSettingsStore().changeCurrentPipelineSetting({ confidence: value }, false)"
+    />
+  </div>
+</template>