Regression updates

[RituYadav92]

1. Added Biomasster regression code in Upernet.py and updated segmentors/init to include the regression single and multitemporal decoder classes.
2. Fixed regression evaluation in evaluator.py [also bug RegEvaluator doesn't work in eval only mode #31]
3. Added regression dataset config: biomassters.yaml
4. Added single and multi-temporal regression configuration files in segmentor: reg_upernet.yaml, reg_upernet_mt.yaml
5. updated biomassters augmentation file, regression_default.yaml
6. fixed NormalizedMinMax function
7. Added random.seed(42) to enable comparison between two fractional trainings (example 10% train labels) by ensuring they are trained on the same data points in each run.