Files whose names occur in the content are attached regardless if they are part of an image syntax

The importer simply iterates over all files in the same namespace as the page and simply checks if the filename occurs in the page's content. This has three problems:

1. The filename might occur outside of a media syntax or could be a substring of another media file and thus the attachment might not be used at all.
2. DokuWiki applies cleaning rules to all ids including media files, see also the deaccent configuration option. For example ä is changed to ae in the default configuration, everything is converted to lowercase and spaces are converted to _. This means that a filename might not occur in the text but could still be referenced.
3. The filename might be encoded, see DOKUWIKI-33

The second problem is unlikely as most media references should be inserted through the media manager that will insert the cleaned name.