Merge load / resize / cache to optimize data loading efficiency for semantic segmentation

[goodsong81]

Summary

[Changes]

• Extend Merge load / resize / cache to optimize data loading efficiency for detection & instance segmentation #2453 to semantic segmentation models
  (all models are sharing the same data pipeline for base incremental learning recipe)
• Enable GT mask caching (after polygon -> mask computation & resizing)

[Results (sampled)]

• kvasir_seg/100 (500x500 ~ 1Kx1K) / LiteHRNet_s_mod2 (512x512) / default setting (num_workers=2)

Branch	Size	Cache	Data time	Iter time	E2E time	Accuracy
develop	512x512	off	0.0245	0.1450	44.32	0.8517
	256x256	off	0.0066	0.1104	45.54	0.8637
	256x256	95M/2G	0.0050	0.1078	35.86	0.8523
feat/load-resize-cache-seg	256x256	off	0.0082	0.1127	37.31	0.8214
	256x256	30M/2G	0.0059	0.1056	37.54	0.8567

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]

LGTM