Shepherd-data

- fix workflows
- latest pydantic-update 2.4 brings some speed improvement for sheep (<10%)

- update fixtures with real testbed-data
- refine testbed-models
- objects with active=False can't be used in experiment
- target.id is now used as selector and is also encouraged to be dynamic to shape the testbed (see layout-map in doc)
- integrate gpio-decoder into reader
- .gpio_to_waveforms()
- .gpio_to_uart()
- rework GitHub workflow to all release and publish by tagging a commit

- uart-decoder: baudrate-detection more robust
- shpModel: change export to make sure set-parameters get exported
- generators: enable rec of uart
- refactor verbosity-system
- GitHubWorkflow refactoring
- add waveform-extraction for shepherd-data - cmd: extract-uart

- breaking change: cores BaseReader and BaseWriter just become Reader & Writer

- allow negative values when converting raw to SI-units
- lift restriction for pandas < v2
- update dependencies and fix incompats & warnings
- warn when files are not overwritten
- create data-generators for the testbed (./extra)
- improve file-handling
- bugfixes for task-handling
- update documentation
- update IPv4-representation
- extend model with more representation-types (yaml, str, dict, ...)

- further fixes for datetimes
- make shpModel even more like a dict
- fix name-collisions
- add CapeData from Sheep
- extend User.active
- add output_paths to task-sets and make them compatible with prepare_tasks() & extract_tasks()