Pytorchlabflow

- customisable training and validation loop
- improved performance_plot

**added remark to each experiment**

- if a config is newly initialises remark="New" [Default] however you can change it to other description like
- if you use a existed ppl without trained weight it takes remark="<used_ppl_name>"
- but if you use with trained weights it takes remark="<used_ppl_name>[<best_epoch_of the given ppl>]"

**corrected use_ppl**

- **improved performance_plot**
adding legend in single experiment plot
- **PipeLine.train( adding patient )**
set pateint value to stop training if there are no improvement in validation loss.
**bug fixed in use_ppl, multi_train, archive, delete**
some syntax errors have been fixed

Only update in **performance_plot**
- **Easy comparison of experiments:** The performance_plot function now supports plotting performance metrics for multiple pipelines.
- **Customizable Plot Size:** Added figsize parameter to control the plot’s aspect ratio.

**Full Changelog**: https://github.com/BBEK-Anand/PyTorchLabFlow/compare/0.1.8.5...0.1.8.6

**Full Changelog**: https://github.com/BBEK-Anand/PyTorchLabFlow/compare/0.1.8.5...0.1.8.6

Access collate function of dataset class which should be defined as static method.