How to handle timeouts from tools and LLM-providers?

[jruokola]

There isn't a clear guidance in documentation how to handle timeouts from LLM providers and tools a best practice example would be welcome

[fbiebl]

Maybe this helps

#1086 (comment)