The Chainlink documentation specifies that you should check that a request for a given
_requestId exists before continuing.
I'm curious to further understand why this is, is it wrong to assume that the Oracle should never attempt to fulfil a request that has never been instantiated? For reference, I have reviewed the best practices section of the documentation and all attached code snippets do not follow this behaviour.
Whilst developing the contract I am working on, I noticed this and wanted to gain a better understanding as to why the documentation validates a given request exists before fulfilling it.