Provider Exception Failure Analyis abstraction to provider better error messages to users

[asaikali]

Since Spring AI integrates with numerous vector stores and models Exceptions are going to happen and users will struggle to figure out why the exception is occurring. Sprig Boot has the concept of Exception failure analyzer by introducing a failure analyzer abstraction into the spring-ai core we can enable model providers, vector store providers and advisor implementers the ability to offer more friendly error messages that will guide users towards a quick resolution of the issues and reduce the volume of Q&A issues raised.

Examples of users raising issues that can benefit from a failure analyzer.

• Conversion Activation Error (Aggregation Error org.springframework.web.reactive.function.client.WebClientResponseException: 413 Payload Too Large from POST https://api.groq.com/openai/v1/chat/completions) #1675