The audio processing delay calculation in the `process_audio_from_buffer` function has been updated. The duration of the audio processing is now compared to the remaining delay, and if it exceeds the delay, the delay is increased by 100 milliseconds. Otherwise, the delay is decreased by 100 milliseconds.

refactor: Update model information retrieval

The code for retrieving model information in the `cleanstream_properties` function has been updated. The loop that adds models to the list now uses the `models_info()` function instead of the `models_info` map directly.

refactor: Update whisper model update logic

The logic for updating the whisper model in the `update_whisper_model` function has been updated. The `models_info()` function is now used to check if the new model path exists in the `models_info` map. Additionally, the `models_info().at(new_model_path)` function is used to retrieve the model information.

refactor: Update model downloader types

The `models_info` map has been changed to a function `models_info()` that returns a constant reference to the map. The `get_sorted_models_info` function has also been added to retrieve a sorted vector of model information.

refactor: Update prebuilt whispercpp version

The prebuilt whispercpp version in the `BuildWhispercpp.cmake` file has been updated to version 0.0.7. The download URL has also been updated to the new repository location.