OTX can't train after clean install because of incorrect package version

[Daankrol]

Describe the bug

After performing a clean install with pip install otx[full] OTX fails to run training:

otx train --data_root cityscapes --task INSTANCE_SEGMENTATION
Usage: otx [-h] [-v] {install,find,train,test,predict,export,optimize,explain} ...
error: Parser key "model":
  Problem with given class_path 'otx.algo.instance_segmentation.maskrcnn.MaskRCNNResNet50':
    No module named 'pytorchcv.models.model_store'

It seems that after installing pytorchcv is upgraded to a higher version by one of the requirements while the 'full' requirements set expects pytorchcv==0.0.67.

Environment:

• OS: Ubuntu 20.04
• Framework version:
• Python version: 3.11
• OpenVINO version:
• CUDA/cuDNN version:
• GPU model and memory: