Apple Silicon Macs via lighter: ONNX detector on the Neural Engine and media engine decode (#24453)

* Run ONNX models on a Mac's Neural Engine through lighter's plugin provider

lighter's lighter.sh/ane device places an ONNX Runtime plugin execution
provider in the container. When it is present, the ONNX session setup
registers it once and opens sessions on its Neural Engine device, the same
place CUDA, ROCm and OpenVINO are chosen, so the onnx detector (and any
model that is not pinned to the CPU) runs there with no configuration. The
hardware probe reports it as an onnx unit.

* docs: hardware decode on an Apple Silicon Mac under lighter

A community section on the video decoding page: lighter's lighter.sh/video
device, hwaccel_args -c:v h264_v4l2m2m, and why the Raspberry Pi presets
decode a single-stream camera in software. The detector docs link to it.

* docs: set the lighter decoder per camera when codecs are mixed

* docs: the ONNX detector on a Mac's Neural Engine under lighter

* Format the Neural Engine provider setup

* Fall back to the default providers when the Neural Engine cannot load a model

* Apple Silicon ffmpeg presets for lighter's media engine, recommended when it is present

docs/docs/configuration/hardware_acceleration_video.md

@@ -43,6 +43,10 @@ Frigate supports presets for optimal hardware accelerated video decoding:
 
 - [RKNN](#rockchip-platform): Frigate can utilize the media engine in RockChip SOCs to accelerate video decoding.
 
+**Apple Silicon Mac** <CommunityBadge />
+
+- [lighter](#apple-silicon-mac-lighter): Frigate can utilize the media engine in Apple Silicon Macs to accelerate video decoding, when running under the lighter container runtime.
+
 **Other Hardware**
 
 Depending on your system, these presets may not be compatible, and you may need to use manual hwaccel args to take advantage of your hardware. More information on hardware accelerated decoding for ffmpeg can be found here: https://trac.ffmpeg.org/wiki/HWAccelIntro
@@ -533,3 +537,35 @@ output_args:
 Make sure that your SoC supports hardware acceleration for your input stream and your input stream is h264 encoding. For example, if your camera streams with h264 encoding, your SoC must be able to de- and encode with it. If you are unsure whether your SoC meets the requirements, take a look at the datasheet.
 
 :::
+
+## Apple Silicon Mac (lighter)
+
+[lighter](https://github.com/fieldwork-ai/lighter) is an open-source container runtime for macOS. It gives a container the Mac's media engine as a standard V4L2 decoder, backed by VideoToolbox, so Frigate decodes H.264 and H.265 streams in hardware with the ffmpeg it already ships. It works on M1 and newer Macs with lighter 0.9.2 or newer.
+
+Give the container the video device. With Docker Compose:
+
+```yaml {4-5}
+services:
+  frigate:
+    ...
+    devices:
+      - lighter.sh/video=all
+```
+
+Or with `docker run`, add `--device lighter.sh/video=all`.
+
+Then set the preset for the codec your cameras stream. The decoder is specific to the codec, so if your cameras mix H.264 and H.265, set the preset for the most common codec globally and override it on the other cameras:
+
+```yaml
+ffmpeg:
+  hwaccel_args: preset-apple-silicon-h264
+
+cameras:
+  garage: # an H.265 camera
+    ffmpeg:
+      hwaccel_args: preset-apple-silicon-h265
+```
+
+The presets decode on the media engine and encode the Birdseye restream and timelapses there too. Scaling to the detect resolution runs on the CPU, as ffmpeg's V4L2 decoders cannot scale.
+
+lighter can also run object detection on the Mac's Neural Engine; see [Apple Neural Engine (lighter)](object_detectors.md#apple-neural-engine-lighter).