Improve `get_model` function to match model names using normalized strings.
Add dash-separated and character-removal normalization strategies, and use
`str.translate()` for efficient string transformations.

- Add translation tables for dash-separated and character-removal normalization
- Support matching models with different separator styles (colons, dashes, dots, underscores)
- Update model data from OpenRouter API (pricing, context lengths, new models)
- Add parametrized tests for various model name formats
- Bump version to 0.7.0 and update dependencies