Many reasons:
* to allow subtrees to be grafted more easily
* to keep the tags for an entry clean by removing structural tags and
leaving the semantic tags
* to avoid the duplication of expressing the same relationship through
both the tags and list fields
1. Switch from using the text field of lists for storing the associated
filter to using the field `toc-list-filter` (to make it harder to
accidentally parse the text of an ordinary tiddler as a filter)
2. Fix several bugs
The exclude filter `+[tag[intro]]` will produce a document that only
includes the paragraphs with the tag “intro”. These are derived from
the paragraphs in the original document with the CSS class “intro”.
