Comment 2 for bug 1483132

The review [1] doesn't waste a word about the encoding, so I would assume that this was unintentional. I struggle with myself to decide if the private SSH key of a keypair should be seen as an interface which Nova exposes for consumption for third party tools. From a security point of view, the stricter DER seems to fit better our needs, so I would assume that a rollback of [1] would be justifiable. Unfortunately I'm no expert here and I need another opinion.

[1] https://review.openstack.org/157931