AmazonBedrockGenerator shouldn't load tokenizer if truncate is set to False.

[hypathia]

Is your feature request related to a problem? Please describe.
DefaultPromptHandler downloads unnecessary GPT2 tokenizer to a read-only filesystem (eg. AWS Lambda), when truncate argument is set to False.

Check

haystack-core-integrations/integrations/amazon_bedrock/src/haystack_integrations/components/generators/amazon_bedrock/generator.py

Lines 148 to 156 in cf52ce9

	# Truncate prompt if prompt tokens > model_max_length-max_length
	# (max_length is the length of the generated text)
	# we use GPT2 tokenizer which will likely provide good token count approximation
	self.prompt_handler = DefaultPromptHandler(
	tokenizer="gpt2",
	model_max_length=model_max_length,
	max_length=self.max_length or 100,
	)

Describe the solution you'd like
If truncate is set to false, GPT2 tokenizer for token estimation shouldn't be loaded.

Describe alternatives you've considered
Simply place a condition to only load the tokenizer if truncate is set to true.

        # Truncate prompt if prompt tokens > model_max_length-max_length
        # (max_length is the length of the generated text)
        # we use GPT2 tokenizer which will likely provide good token count approximation
--->    if self.truncate:
            self.prompt_handler = DefaultPromptHandler(
                tokenizer="gpt2",
                model_max_length=model_max_length,
                max_length=self.max_length or 100,
            )