background-video doesn't honor imagesdir attribute like it should

[obilodeau]

Investigating what I thought was a problem with Asciidoctor, I realize that we are badly loading background-video local files.

Asciidoctor uses the media_uri method to lookup video files and we used it in the video template but for background-video this wasn't used.

This PR fixes that behavior.

Now, is this a backward-incompatible fix? It could break existing decks that relied on the wrong behavior. However, it was obviously broken and didn't follow what an Asciidoctor/AsciiDoc user would expect. I'm open to make this change part of a feature release but I would find it intense to bump a major version because of this. Guidance needed here.

Note

• Title slide background-video is already correctly implemented
• The first commit just exhibit the bug (I want to see it in travis), I have a local fix read to be pushed fix pushed

[ggrossetie]

Now, is this a backward-incompatible fix? It could break existing decks that relied on the wrong behavior. However, it was obviously broken and didn't follow what an Asciidoctor/AsciiDoc user would expect. I'm open to make this change part of a feature release but I would find it intense to bump a major version because of this. Guidance needed here.

It reminds me of: https://xkcd.com/1172/ (to a certain extend)
The behavior was defined by Asciidoctor and we should resolve images, audios and videos from the imagesdir directory, so we are fixing a bug.

Having said that you should put emphasis on the release note but I don't think we should do a major release. Otherwise, potentially every changes/bug fixes could potential break an existing workflow and thus requiring a major release.