Expanding model support for MyxMatch

[salma-remyx]

This PR helps expand the available models to evaluate through MyxMatch. We introduce additional validators to confirm availability offline before job submission.

Test

To test this latest change, you can try running a new myxmatch evaluation:

from remyxai.client.myxboard import MyxBoard
from remyxai.api.evaluations import (
    EvaluationTask,
    BenchmarkTask
    )
from remyxai.client.remyx_client import RemyxAPI

myx_board_name = "small-llm-comparison-media_example"
model_ids = ["microsoft/Phi-3-mini-4k-instruct", "Qwen/Qwen2-1.5B"]
myx_board = MyxBoard(model_repo_ids=model_ids, name=myx_board_name)

tasks = [EvaluationTask.MYXMATCH]
prompt = "You are a media analyst. Objective: Analyze the media coverage. Phase 1: Begin analysis."

remyx_api = RemyxAPI()
remyx_api.evaluate(myx_board, tasks, prompt=prompt)
>> Starting evaluation...

>>Evaluations are done! You can now view the results.
results = myx_board.get_results()

results
>>
{'myxmatch': [{'model': 'microsoft/Phi-3-mini-4k-instruct',
   'rank': 1,
   'prompt': 'You are a media analyst. Objective: Analyze the media coverage. Phase 1: Begin analysis.'},
  {'model': 'Qwen/Qwen2-1.5B',
   'rank': 2,
   'prompt': 'You are a media analyst. Objective: Analyze the media coverage. Phase 1: Begin analysis.'}]}

[smellslikeml]

parsing this way will fail for so many model ids, we need something more robust:
https://stackoverflow.com/questions/68086929/how-to-get-the-size-of-a-hugging-face-pretrained-model

[smellslikeml]

model size validation needs something more reliable

[salma-remyx]

model size validation needs something more reliable

noted in the remyx repo