The difference between this split, and the old, is that everything is
now clearly under output.

The files
- format
- gather
- templates
- output
have now all been merged into `output`.

The difference between `format` and `output` was allways unclear.
`gather` could be separate, but it was still so tightly linked to the
output.

`templates` could have been kept, but it started depending on types from
`output`, making it easier to merge (at least of now).

All state held in that class is apparently now moved to the
ParserCombinator which uses the serialized output

Possibly change all methods in the class to be class/static methods
instead in the future.

Change name from 'Items' to 'Parser'
====================================
This one was a long tim comming. Items was a transient name due to the
parser combinator library willing itself into existance without my
consent.

Made a bunch of classes private
===============================
The ParseDirective classes only present to enable a "primitive" global
parser are now marked private. There is zero reason to actually
instanciate them more than once, but they still need to be classes due
to the API. The corresponding global "variables" now have attached
documentation and get exported by sphinx.

Distribute documentation
========================
Previously almost all documentation for the parser combinators lived in
the header of the file, but is now distributed onto the declarations.
The previous version was mostly due to me not understanding how to
export global "variables" to sphinx.

Default logging level should be INFO or WARNING, however, we really
should load this from a configuration file, enviroment variables, and
command line argumentns.

.coverage appears to be a file generated when pytest is run with
coverage reporting enabled. It's nontheless a binary file, and should be
ignored.

Previously, all possible parse instances of a string were shown, which
included the unquoted, single quoted, and double quoted versions,
meaning that "always" was shown as:

    string(([ws] & ('always'
		    | ("'" & ['a', 'l', 'w', 'a', 'y', 's'] & "'")
		    | ('"' & [rich_char(c='a'), rich_char(c='l'), rich_char(c='w'), rich_char(c='a'), rich_char(c='y'), rich_char(c='s')] & '"'))))

This changes the output to:

    string([frequency])

Extend DocStringTag to include "all" known types, along with document
how each of them behaves. This also forced a few changes to its
behaviour, which is reflected in the updated tests. The biggest change
is that we now barely test for valid "tag_name" values, since a valid
tag should always have a valid (albeit unknown) tag name.

Errors are now created through a factory, embedding information about
WHERE the error comes from in the source.

Also add a clearer indicator to the error output.

As a steping stone, the raw match data was emited to the frontend. This
just properly renders it.

If `parse_puppet` crashed, instead format the the raw code, adding an
error marker on the erroring character if possible. The error message is
also added.