feat: add run_finetune core agent action example

[ZILECAO]

This PR adds the RunFinetuneAction tool call for the Hyperbolic agent. When prompted to run a fine tuning task with a specific model (eg. unsloth/mistral-7b-v0.3-bnb-4bit), the Hyperbolic agent will sync an example fine tuning folder with all required materials to a remote GPU instance rented from Hyperbolic. It will then try to run an example fine tune task. The main purpose of this integration is to quickly get a fine tuning environment set up for GPU users. It's not in a state where a user can freely fine tune solely by prompting the chat agent interface. After the initial run_finetune tool call, it is advised for a user to track their fine tune progress by SSHing into their remote GPU, and modifying with training data files and fine tuning scripts directly there, although more actions could come soon.

The example folder "finetune_example" contains:

1. example unsloth fine tune script (finetune.py)
2. example training data (training_data.jsonl)
3. example inference call using the newly finetuned model (test_inference.py)
4. strict pip dependency file (requirements.txt)

A demo video can be found here: https://www.loom.com/share/13dfa667db9f496188df284cb15c392b?sid=397e07f7-fb69-472e-8f93-29abea759ce8

and more quickstart info can be found in the README.md inside the finetune_example folder

[aabdel0181]

tested this out earlier -- works great locally on WSL 2.