Content of text nodes is not escaped

When writing normal Elm, one does not have to do much escaping to avoid xss since, for example, `Html.text` just creates a text node, which cannot be interpreted as containing any tags.

When converting to html though, this library should ideally escape the contents of text nodes, so a view like

``` Elm
view : Html msg
view =
Html.text "//do evil stuff"
```

does not generate malicious html files.

I realize this project may be superceded by a native Elm workflow in 0.19, but until then I feel this is still a security concern.

The reported entry point is Elm.text during conversion to HTML; start by reproducing the supplied view and tracing where its text content is serialized. Done means the generated HTML escapes the script-like text so it remains a text node, with no tags interpreted.

Written by the indexing model from the issue text.