`nodes.Apax` retrain issues

[PythonFZ]

• [ ] no information that the training has been resumed from the train.log file
• [ ] the log.csv entries are purged when rerunning the Node.
• [ ] use self.state.restarted and bump minimum ZnTrack version to ensure that it can be handled: https://github.com/apax-hub/apax/blob/479320502a0c9f520a2b704be96abbd4474d3f48/apax/nodes/model.py#L92-L93
• [ ] test that calling zntrack run apax.nodes.model.Apax --name Apax twice when the first run timed out / was killed yields the expected results
• [ ] adapt apax log level