Depthai

Features
* **ImageAlign** node
* it can align depth to any other sensor on the device - works for ToF too.
* it can align two sensors with each other using intrinsics only, useful for thermal-rgb alignment
* **Cast** node
* Cast `NNData` message to `ImgFrame`
* Useful in case apps need to use outputs from `NeuralNetwork` node to be fed into nodes that only accept `ImgFrame`
* **Full ToF support**
* Running live at 30 FPS
* Measuring range of 20cm - 6m
* < 1% error across the range
* Support for ToF in spatial nodes
* Add an option to limit bandwidth over XLink `setXLinkRateLimit(int maxRateBytesPerSecond, int burstSize, int delayUs)`

Stability improvements and bug fixes
* Improved PoE stability on reboots - eliminate the case where powercycle of the device was sometimes needed
* Improved runtime stability of heavy pipelines by increasing priority of the cameras in the NoC
* Improved ImageManip stability
* Improved XLink communication to be able to detect memory corruption and avoid it
* Fix a bug where stereo rectification was inaccurate when the calibration data didn't contain direct link between the two inputs
* Relevant for custom setups on FFC devices

Misc
* Improve numerical stability of the rectification algorithm
* Improves stereo quality on wide FOV sensors

Features
- Added `DeviceBootloader::getFlashedVersion` to retrieve the bootloader version that is flashed on device.

Bug Fixes
- Fixed fsync on OAK-D-SR.
- Fixed boot issue on OAK-D-SR-POE and OAK-T.
- Fixed compilation in same cases, because of problems with `jsoncpp`. (https://github.com/luxonis/depthai-core/pull/980)

Misc

Features
- Added `DeviceBootloader::getFlashedVersion` to retrieve the bootloader version that is flashed on device.

Bug fixes
- Fixed fsync on OAK-D-SR.
- Mitigated boot issue on OAK-D-SR-POE and OAK-T.
- Fixed compilation in same cases, because of problems with `jsoncpp`. (https://github.com/luxonis/depthai-core/pull/980)

Misc
- Added thermal neural network example (`examples/NeuralNetwork/thermal_nnet.py`).

Features
* **PointCloud** node along with the PointCloudData message
* On device pointcloud generation
* Native conversion to PCL message type - similar to `getCvFrame()`
* Support for the **OAK Thermal**
* New `setManualFocusRaw()` API with more granularity for setting manual focus position
* New `getStereoPairs()` and `getAvailableStereoPairs()` API to allow more device agnostic programming
* Add the `intensity` output to the ToF node
* Add the crop sensor configs in `getConnectedCameraFeatures()` now contain how each of the resolutions is cropped from the native sensor resolution

Bug fixes
* Bug fix for `v2.24.0` regression with flashing the bootloader [link](https://github.com/luxonis/depthai-python/issues/953)
* Bug fix in message groups for large resolution frames when going through `XLinkOut` node
* Bug fix to apply AE limit during AF lens movement (before the fix, the AE limit wasn't applied during active AF movements)
* Bug fix for grayscale `VideoEncoder` node with 2 inputs of different sizes

Misc
* Fix timings for IMX214 4K resolution, which were slightly off

Features
* **PointCloud** node along with the PointCloudData message
* On device pointcloud generation
* Native conversion to PCL message type - similar to `getCvFrame()`
* Support for the **OAK Thermal**
* New `setManualFocusRaw()` API with more granularity for setting manual focus position
* New `getStereoPairs()` and `getAvailableStereoPairs()` API to allow more device agnostic programming
* Add the `intensity` output to the ToF node
* Add the crop sensor configs in `getConnectedCameraFeatures()` now contain how each of the resolutions is cropped from the native sensor resolution
Python only features
* Added `depthai` binary
* Useful to quickly list out devices, run `cam_test`, etc.
* Added Python 3.12 wheels
* Added 32 bit wheels for `bookworm` OS on RPI (Python 3.11)

Bug fixes
* Bug fix for `v2.24.0` regression with flashing the bootloader [link](https://github.com/luxonis/depthai-python/issues/953)
* Bug fix in message groups for large resolution frames when going through `XLinkOut` node
* Bug fix to apply AE limit during AF lens movement (before the fix, the AE limit wasn't applied during active AF movements)
* Bug fix for grayscale `VideoEncoder` node with 2 inputs of different sizes

Misc
* Fix timings for IMX214 4K resolution, which were slightly off

Features
* New nodes and messages:
* **Sync node** - syncs multiple inputs based on the timestamp, outputs a message group message
* **Demux node** - demultiplexes message group in multiple messages
* **Message group message** - a new message that can contain a map of arbitrary depthai messages, it's the output of the sync node and input to the demux node
* **Encoded frame message** - a new message specialized for encoded frames
* New output for the VideoEncoder node (`out`) for the encoded frame message
* Automatic crash dump retrieval for firmware crashes
* Added `setIrFloodLightIntensity` and `setIrLaserDotProjectorIntensity` methods for setting the intensity normalized between 0 and 1
* Added `getConnectionInterfaces` method to retrieve the list of available interfaces on a device
* Added an option to cap maximum time for exposure when using auto exposure with `setAutoExposureLimit`
* Initial integration for `IMX283` and `IMX462`
* Improve time-syncing between the host and device to achieve sub 300 us offset
* Improved max FPS and image quality under low light for `OV9282` and `OV9782` https://github.com/luxonis/depthai-core/pull/926 ,new ranges per resolution of:
- `THE_800_P`: `1.687` .. `129.6` fps
- `THE_720_P`: `1.687` .. `143.1` fps
- `THE_400_P`: `1.687` .. `255.7` fps


Bug fixes
* Avoid overflow for XLink profiling https://github.com/luxonis/depthai-core/pull/933
* Improve XLink stability when using multiple devices https://github.com/luxonis/XLink/pull/73
* Fix a rare bug where the device would hand in the constructor https://github.com/luxonis/depthai-core/pull/922
* Fix a bug where `XLinkIn` didn't work correctly for very small and very large buffers
* Fix a bug for running multiple stereo nodes with a shared input


Misc
* On multi-input NeuralNetworks set the output NNData to the newest input timestamp (previously undefined)
* Add NOC DDR usage reporting on `DEPTHAI_LEVEL=info`