Per the CSOT spec (https://github.com/mongodb/specifications/blob/master/source/client-side-operations-timeout/client-side-operations-timeout.md#errors) – "external" timeouts (like server, network sockets, HTTP timeouts, etc) ought to be represented by the new TimeoutError class (but only when timeout_ms is present).

I propose the following to implement this:

• Refactor the behavior of OperationFailure (the current error class for server-side timeouts) into a module (e.g. OperationFailureBehavior).
• Subclass TimeoutError with a new ServerTimeoutError, which implements this new OperationFailureBehavior module.
• Change our error handling code so that it looks for OperationFailureBehavior instead of OperationFailure, allowing both operation failures and server timeouts to be handled by the current rescue handlers.
• Allow an OperationFailure to be morphed into a ServerTimeoutError in such a way that the new error still accurately represents the original error, while also standing for a TimeoutError .
• Ideally, we only use ServerTimeoutError when timeoutMS was provided to the original operation. To make this happen, we need to pass context to