feat: read from litellm defaults for context window limits

[shreyashankar]

Purpose

This PR reads from litellm's model_cost defaults instead of using hardcoded values for context limits. The LM class now automatically derives max_ctx_len and max_tokens from litellm's model information when these parameters are not explicitly provided.

Test Plan

Tested a lotus tutorial and my own scripts with long documents with the gpt-4.1-mini model (which has a 1M context window), to ensure we weren't running into context window issues.

Test Results

N/A

(Optional) Documentation Update

N/A

Type of Change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update
• Performance improvement
• Refactoring (no functional changes)

Checklist

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code, updating docstrings
• I have made corresponding changes to the documentation
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes

mypy was failing before making the change; I didn't touch files outside of lm.py.

BEFORE SUBMITTING, PLEASE READ https://github.com/lotus-data/lotus/blob/main/CONTRIBUTING.md
anything written below this line will be removed by GitHub Actions