EhÂ… thats more an example of ignorantly modifying code to silence tool warnings, not really much of an example of the tricky implications of cryptography. At most you can say about the debian openssh example is that it shows that security is often an invisible property, but that isn't a crypto specific point... and you can argue that crypto should be left to the cryptonauts but security really must be every developers problem.
The mention of RC4 in WEP in the article makes a better example of the special challenges posed by cryptography, or perhaps the old watermarking attacks against pure CBC dmcrypt volumes prior to the introduction of ESSIV and LRW... the point that you can use the primitives correctly but still produce something insecure because of non-obvious (and sometimes highly mathematical) properties of the cryptographic components.