feat: add only to booleanFooter

[fyInALT]

Relates to:

Risks

medium

Background

When use llama3.1:8b as OLLAMA_MODEL, when use booleanFooter, it will return not just YES or NO, like:

Use context:

   {"context":"# Task: Decide if the recent messages should be processed for token recommendations.

    Look for messages that:

    ...  

    Should the messages be processed for recommendations? Respond with a YES or a NO."} 

and will return:

   {"response":"YES 

There are several instances in this conversation where users ask to create tokens, send Solana coins, and discuss token-related topics. The keyword \"token\" is mentioned multiple times, indicating that these conversations should be processed for token recommendation."} 

It will make the match in parseBooleanFromText return null, then make generateTrueOrFalse call blocked by

while (true) {
    try {
      const response = await generateText({
        stop,
        runtime,
        context,
        modelClass
      });
      const parsedResponse = parseBooleanFromText(response.trim());
      if (parsedResponse !== null) {
        return parsedResponse;
      }
    } catch (error) {
      elizaLogger.error("Error in generateTrueOrFalse:", error);
    }
}

So that the eliza will keep retry to got the response.

It can fixed by add a only in booleanFooter. Then it will only return YES or NO

What does this PR do?

This PR change booleanFooter to:

Respond with only a YES or a NO.

What kind of change is this?

Bug fixes

Documentation changes needed?

No

Testing

Where should a reviewer start?

Detailed testing steps