page: read file during article object initialisation

Derive parts (title, body) from full contents instead of reading piecemeal.
Assume most pages are small enough so added overhead should be minimal,
adding support for titles after the first line.