Refactor DetCon logic & Update SupCon for seg recipes

[sungchul2]

Summary

This PR includes

• enable to use two views per image instead of single view
• update lr warmup iters in supcon recipe
• create DetConHead including detcon loss and predictor
• update configs & docstrings
• add unit test

Through these changes, we can expect improved mDice (~+4.96) and reduced training time (~0.68x) compared with base supervised learning.

How to test

tests/unit/algorithms/segmentation/adapters/mmseg/models/heads/test_detcon_head.py

tests/unit/algorithms/segmentation/adapters/mmseg/models/segmentors/test_detcon.py

tests/unit/algorithms/segmentation/adapters/mmseg/models/losses/test_detcon_loss.py

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

[sungmanc]

I changed the title of this PR, we're not using [FIX], [ENHANCE] prefix anymore.

[sungmanc]

LGTM, it showed quite improvement for SupCon algorithm. How about adding this changes to CHANGELOG ?

[jaegukhyun]

Ah, there is test failure, could you check it?

[sungchul2]

@sungmanc @jaegukhyun
I updated CHANGELOG and self-sl recipes to resolve integration test errors.
Could you review it again?

[supersoob]

LGTM