Strip HTML out of descriptions.

[rossjones]

We have some descriptions that contain HTML, it's rendered safely but we probably should remove it.

Related: should we auto-link anything that looks like a link?

[gtdata]

@rossjones not sure what this means - will you show me pls?

[rossjones]

@gtdata http://alpha-find-data.herokuapp.com/dataset/leeds-inspired-api is a good example, the dataset was harvested from a system that uses HTML for description, and so they get pulled through. In this case it's just a para tag, we could remove those. Some datasets though (can't easily find one atm) use HTML for formatting and lists.

The existing system allows Markdown, which is stored as markdown, but rendered as HTML. But not all of the things we harvest, do this.

[rossjones]

I think stripping HTML out of harvested datasets makes sense.

When we import legacy stuff, we can also strip it there.