Object Detection - Remaining Items

### Required

#### Testing
- [x] Add OD and DataSet to Test Matrix for Testing
- [x] Figure out if there is actually an issue causing Test/Validate split of 0.1 not to work?

#### Code Gen
- [x] Add Box Resizing in Code Behind and Generated Projects
- Protype Code Here: https://github.com/dotnet/machinelearning-modelbuilder/issues/2610#issuecomment-1528945535
- [x] Add VOTT data loading to *.training.cs so that users have everything they need to re-train
- Protype Code Here: https://github.com/dotnet/machinelearning-modelbuilder/issues/2610#issuecomment-1528945535
- [x] Add COCO data loading to *.training.cs so that users have everything they need to re-train
- [x] 800 and 600 should be generated as constants
- [x] Fix formatting of *.training.cs

#### ML.NET and AutoML
- [x] Update TorchSharp
- ML.NET https://github.com/dotnet/machinelearning/pull/6636
- Model Builder https://github.com/dotnet/machinelearning-tools/pull/1751
- Generated projects will fail until we pull in new version

#### Other
- [x] Update training to use a lower filter threshold for boxes - 0.5 is too high. The user can always filter the results more like they do for the azure model.
- [x] Default epoch to something larger than 1
- [ ] Add document/link/tutorial for coco option

### Nice to Have

- [x] Add progress information
- [ ] Work on getting box resize transform into ML.NET
- https://github.com/dotnet/machinelearning/issues/6629
- [ ] Support TSV Input Format
- https://github.com/michaelgsharp/machinelearning/blob/main/test/data/images/object-detection/fruit-detection-ten.tsv
- [x] Support COCO Input Format
- https://github.com/dotnet/machinelearning-tools/pull/1752
- [ ] Continue training
#### Training Options
- [x] Add Training Option in UI for Training Image Size (Default: 800x600)
- [x] Add Training Option in UI for max epoch (Default: 5)

Beginne damit, die nicht abgehakten Punkte in diesem Issue zu prüfen, insbesondere die Anfrage zur COCO-Dokumentation, die Unterstützung für TSV-Eingaben, die Arbeiten zur Größenänderung von Boxen und das Fortsetzen des Trainings. Lies das verlinkte ML.NET-Issue und die referenzierten Beispiele für Trainingsprojekte, bevor du den Umfang festlegst; „fertig“ sollte bedeuten, dass ein klar abgegrenzter Punkt implementiert oder dokumentiert wurde und der zugehörige Trainingsablauf verifiziert ist.

Vom Indexierungsmodell aus dem Issue-Text verfasst.