Options to override loggers and progress bar for `run_inference`

[tom21100227]

Currently sleap_nn.predict:run_inference uses logger from loguru and have a built in tdqm style progress bar. It would be nice if the function accepts an external logger and/or a progress call back function. When parallelizing inference jobs having an override-able logger would make a lot of things easier for observability. The progress bar might clogs up STDOUT with how some logs handle carriage-return updates.

I am imagining something similar to this:

def run_inference(
    ...,
    logger: logger = None, # option to override the built in logger
    progress_bar: bool = True # whether to enable the progress bar or not. 
    progress_callback: func = None # a progress function that can be used to replace the progress bar in the shape of func(progress: int). 
)