Tasks: Add Subtasks under Reinforcement Learning

[MrShahzebKhoso]

Currently, Reinforcement Learning is listed as one broad category. I suggest adding more subtasks under this category (similar to how Computer Vision etc. is structured). Example tasks could be:

• Single-Agent RL
• Multi-Agent RL
• Model-Based RL
• Policy-Based RL
• Value-Based RL
• Safe RL
• RLHF
• Imitation / Offline RL

This would improve discoverability, align RL with other domains, and make the tasks page more useful. I’d be happy to contribute the initial definitions and metadata if maintainers agree.

Would love your feedback before I prepare the PR.

Thanks!