[tinker] SkyRL tinkerification tracker

[SumanthRH]

Summary

Adding this parent issue tracking all the outstanding issues with the tinker API server in SkyRL

SkyRL-Train Backend

• Support PPO loss with Tinker
  - Support critic model in SkyRLTrainBackend
• KL Penalty is not supported
  - Support ref model in SkyRLTrainBackend
• Support logprobs with the sample() API.
  - Provide an example for using truncated importance sampling with the tinker API server.
  - assigned to @tamoghnokandar
• Better validation for configuration parameters to the SkyRLTrainBackend [tinker] Implement better validation for CLI parameters with the skyrl-train backend #1279
• Set LR outside of workers in skyrl-train backend: [tinkerification] Remove scheduler from inside the workers and default to use set_lr() and get_lr() in training loop #998
• Add sample API to the new inference codepath: [tinker] Sample API in new inference server #1286
• Add renderer to the new inference codepath: [SkyRL][tinker] Add renderer endpoint to inference client #1288

Jax Backend

• Improve the memory handling and reporting to make it easier to see where the memory goes
• Add multimodal support to APIs and add support for a VLM model