[Feature Request]: LLMConfig respect None api key value

[CatBraaain]

What needs to be done?

from crawl4ai.async_configs import LLMConfig
print(LLMConfig(provider="openrouter/deepseek/deepseek-r1:free").__dict__)

This goes {'provider': 'openrouter/deepseek/deepseek-r1:free', 'api_token': 'no-token', 'base_url': None}
But i need {'provider': 'openrouter/deepseek/deepseek-r1:free', 'api_token': None, 'base_url': None}
Because api_token must be None for LiteLLM resolving api key automatically from reading .env file.

What problem does this solve?

This feature solve the problem LLMConfig not supporting all providers/models of LiteLLM supported.

Target users/beneficiaries

Users

Current alternatives/workarounds

Overwrite api_token prop after constructor

llm_config = LLMConfig(provider="gemini/gemini-2.0-flash-exp", api_token = None)
# {'provider': 'openrouter/deepseek/deepseek-r1:free', 'api_token': 'no-token', 'base_url': None}
llm_config.api_token = None
# {'provider': 'openrouter/deepseek/deepseek-r1:free', 'api_token': None, 'base_url': None}

Proposed approach

Make api key argument Noneable.
Even passing only provider/model name to LiteLLM, LiteLLM read .env file and set api key automatically.
This is useful to use fallbacks.

using fallbacks like this

# test.py
    LLMExtractionStrategy(
        instruction="Extract title of the page",
        llm_config=LLMConfig(provider="gemini/gemini-2.0-flash-exp"),
        extra_args={
            "fallbacks": [
                "openrouter/deepseek/deepseek-r1:free",
            ]
        },
    )

# .env
GEMINI_API_KEY="xxxxxxxxxx"
OPENROUTER_API_KEY="xxxxxxxxxxxxx"