Format span attributes in AI integrations

[antonpirker]

The AI Agents integrations render stringified json-like data in a nice way (make the sub nodes of the data structure collapsible)

In Javascript it comes down to having double quotes in a string:

• Good: '{"role": "system", "content": "some context"}'
• Bad: "{'role': 'system', 'content': 'some context'}"

Also pydantics model_dump() sometimes returns function or class objects that can not be json serialized so I updated _normalize_data() to make sure everything is converted to a primitive data type, always.

[codecov]

❌ 24 Tests Failed:

Tests completed	Failed	Passed	Skipped
23946	24	23922	2304

View the top 3 failed test(s) by shortest run time

tests.integrations.aws_lambda.test_aws_lambda::test_basic_exception

Stack Traces | 0s run time

.../integrations/aws_lambda/test_aws_lambda.py:70: in test_environment
    LocalLambdaStack.wait_for_stack()
.../integrations/aws_lambda/utils.py:221: in wait_for_stack
    raise TimeoutError(
E   TimeoutError: AWS SAM failed to start within 60 seconds. (Maybe Docker is not running?)

tests.integrations.aws_lambda.test_aws_lambda::test_non_dict_event[event as list of dicts]

Stack Traces | 0s run time

.../integrations/aws_lambda/test_aws_lambda.py:70: in test_environment
    LocalLambdaStack.wait_for_stack()
.../integrations/aws_lambda/utils.py:221: in wait_for_stack
    raise TimeoutError(
E   TimeoutError: AWS SAM failed to start within 60 seconds. (Maybe Docker is not running?)

tests.integrations.aws_lambda.test_aws_lambda::test_trace_continuation

Stack Traces | 0s run time

.../integrations/aws_lambda/test_aws_lambda.py:70: in test_environment
    LocalLambdaStack.wait_for_stack()
.../integrations/aws_lambda/utils.py:221: in wait_for_stack
    raise TimeoutError(
E   TimeoutError: AWS SAM failed to start within 60 seconds. (Maybe Docker is not running?)

To view more test analytics, go to the Test Analytics Dashboard
_{📋 Got 3 mins? Take this short survey to help us improve Test Analytics.}