Hotfix: Reorganize conditional_train code

[chrstnkgn]

What

Reorganize conditional train code, and simplify its config

Why

• We are aware that we should somehow modify the code for conditional training, but it has been delayed as it was not a priority.
• However, as we are preparing for our first release, neater (and well-working) code is needed.
• This issue will cover following parts:
  1. (conditional_train.py) Merge train and trainval function - this format is not necessary anymore as we are not using ray tune for conditional learning, and this will resolve some error that was caused by its complicated format.
  2. (conditiona_train.yaml) Modify its format - As this is not a major option that the users will always take into account, it would be better to place this config somewhere in the lower level. Trimming some unnecessary configs is also needed.

How

• conditional_train.py
  • merge train and trainval
  • remove unnecessary codes
• conditional_train.yaml
  • Modify format

[seoulsky-field]

Great issue! I was thinking when we revise this and how we do that. And I agree this is the best time to do now. Thanks for your efforts. If you have any supports to do this issue, I'll help right away.