Spring-ai AWS BedrockAnthropic3ChatModel token usage metrics are missing

Bug description In the AWS BedrockAnthropic3ChatModel sync call the token usage metrics is not returned. Also bedrock return other very important metric X-Amzn-Bedrock-Invocation-Latency that would be good to include in the spring ai response model

Environment Last M3 version

Steps to reproduce Make a simple call with the BedrockAnthropic3ChatModel

Expected behavior Have all response metrics that bedrock returns

Thanks in advance