Doc typo: rd_kafka_conf_destroy(rk) should be (conf)

[Quuxplusone]

As mentioned in #4100.

[Quuxplusone]

I've rebased the patch. We still haven't resolved what is the intended semantics of fail(...) in this example. Being a C++ chauvinist myself, I prefer to think of fail(...) as doing a throw or longjmp, so that we still need to clean up before calling it. That's also the most useful semantics for documentation purposes, because it gives us a chance to show the reader what they actually need to do to clean up after each of these calls. If we assume fail(...) means abort(), then we don't need to clean up (in this example) but we also can't show the reader what they need to do to clean up (in actual practice).

[cla-assistant[bot]]

All committers have signed the CLA.

[Quuxplusone]

New year's ping!