Add training loss dynamics exportation feature for detection task

[vinnamkim]

Summary

• Ticket no. 93287
• Add training loss dynamics exportation feature for detection task.
• This PR only supports CustomATSS detector. Following up PRs will cover other models.

How to test

I added unit tests about this change.

Checklist

• I have added unit tests to cover my changes.​
• I have added integration tests to cover my changes.​
• I have added e2e tests for validation.
• 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) 2023 Intel Corporation
# SPDX-License-Identifier: Apache-2.0

[codecov-commenter]

Codecov Report

Patch coverage: 95.47% and project coverage change: +0.33 🎉

Comparison is base (63d9d7a) 79.55% compared to head (0a595b6) 79.89%.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #2109      +/-   ##
===========================================
+ Coverage    79.55%   79.89%   +0.33%     
===========================================
  Files          462      464       +2     
  Lines        34124    34325     +201     
===========================================
+ Hits         27149    27424     +275     
+ Misses        6975     6901      -74     

Flag	Coverage Δ
py310	79.89% <95.47%> (+0.33%)	⬆️
py38	?
py39	79.88% <95.47%> (+0.32%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
...ithms/detection/adapters/mmdet/models/loss_dyns.py	77.27% <77.27%> (ø)
otx/core/data/noisy_label_detection/base.py	89.28% <90.00%> (-1.20%)	⬇️
...on/adapters/mmdet/models/heads/custom_atss_head.py	92.39% <95.65%> (+67.05%)	⬆️
...ication/adapters/mmcls/models/classifiers/mixin.py	96.55% <100.00%> (-0.17%)	⬇️
...ithms/detection/adapters/mmdet/datasets/dataset.py	94.36% <100.00%> (+0.12%)	⬆️
...dapters/mmdet/datasets/pipelines/load_pipelines.py	89.36% <100.00%> (+0.23%)	⬆️
...ers/mmdet/models/detectors/custom_atss_detector.py	67.56% <100.00%> (+1.81%)	⬆️
...ters/mmdet/models/detectors/loss_dynamics_mixin.py	100.00% <100.00%> (ø)
otx/algorithms/detection/adapters/mmdet/task.py	85.42% <100.00%> (+1.05%)	⬆️
...hms/detection/adapters/mmdet/utils/config_utils.py	91.82% <100.00%> (-1.96%)	⬇️
... and 3 more

... and 11 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[sungmanc]

LGTM

[jaegukhyun]

Overall codes look good to me. Since I'm not familiar with loss dynamics, I don't closely look at the algorithm in detail. I wish I could review it after introduction meeting. And I want to suggest to move config patch parts to existed patch function and configurer.

[wonjuleee]

LGTM