Improve YOLOv9 performance

[sungchul2]

Summary

This PR includes:

• Update hyperparameters and configurations for performance improvement
• Fix loss weights
• Update export option to align performance of torch and ov IR
• Enable to skip current batch without ground truth
  • This feature is because yolov9 loss cannot calculate loss values without ground truth

**The below table shows F1-score on pothole_medium before and after this update and the score on bdd_medium.

model	before (pothole_medium)	after (pothole_medium)	after (bdd_medium)
v9_s	0.490	0.737 (+0.247)	0.513
v9_m	0.442	0.706 (+0.264)	0.558
v9_c	0.255	0.731 (+0.475)	0.560

**The below table shows F1-score consistency between torch and ov IR models after this update.

model	dataset	type	mAP	F1-score
v9_s	pothole_medium	torch	0.343	0.737
		ov	0.402	0.754
		ov (int8)	0.399	0.748
	bdd_medium	torch	0.142	0.513
		ov	0.182	0.517
		ov (int8)	0.167	0.506
v9_m	pothole_medium	torch	0.335	0.706
		ov	0.368	0.725
		ov (int8)	0.375	0.709
	bdd_medium	torch	0.170	0.558
		ov	0.189	0.556
		ov (int8)	0.190	0.550
v9_c	pothole_medium	torch	0.377	0.731
		ov	0.413	0.747
		ov (int8)	0.436	0.766
	bdd_medium	torch	0.202	0.560
		ov	0.215	0.572
		ov (int8)	0.222	0.560

How to test

Checklist

• I have added unit tests to cover my changes.​
• I have added integration tests to cover my changes.​
• I have ran e2e tests and there is no issues.
• I have added the description of my changes into CHANGELOG in my target branch (e.g., CHANGELOG in develop).​
• I have updated the documentation in my target branch accordingly (e.g., documentation in develop).
• I have linked related issues.

License

• I submit my code changes under the same Apache License that covers the project.
  Feel free to contact the maintainers if that's a concern.
• I have updated the license header for each file (see an example below).

# Copyright (C) 2024 Intel Corporation
# SPDX-License-Identifier: Apache-2.0

[sungchul2]

This PR will be updated more soon, don't merge it.

[kprokofi]

detection tests are passed

[sovrasov]

@kprokofi thanks for arranging the tests. Merging without the queue, since it doesn't work