i.e. Overview is now a mostly pure aggregator again.

The new commit / blip feature is now maintained in `blip.kummerlaender.eu`.

This is the starting point for extending _Overview_ to support various
types of recent updates (i.e. not just blog articles and commits over a
certain message size).

While direct fetching and rendering of markdown-formatted Git commit
messages is an attractive concept, this new approach should be much more
flexible. As a side benefit commit entries can now be edited and
expanded compared to their fixed Git reference. e.g. retroactively
expanding upon older commits or adding example media that doesn't
strictly belong inside the actual commit message.

Fetching external resources during the actual build became problematic
after the last major Nix update (build purity was improved).

The sensible way of doing things in Nix seems to be to declare ALL
build dependencies prior to the actual build -- this includes
e.g. ATOM feeds to be aggregated.

I am probably not going to do the same for the individual repository
feeds as this feature is going to be replaced with a more general
_microblog_-like subpage.

* commit `71a637` in upstream _InputXSLT_ activated namespace comprehension for `InputXSLT:read-file` and `InputXSLT:external-command`
** this caused both the article and the repository feed queries to fail as they are inside the ATOM namespace

* added overview commit count parameter
* added `key` subdomain link

* the first paragraph of the three latest articles of my blog are now displayed on the overview page
* feed is fetched via `curl` in `01_raw/article_feed.xsl`
** as with all commit feeds the article feed is processed further in the `02_augment` level
** the count of articles to be displayed is defined in `00_content/meta.xml` as `article_count`
* CSS was updated to correctly display the article summary

* I just realized that repository specific commit feeds are already provided by both CGit and GitHub
** i.e. I made the mistake of not checking if my current system already provides the required functionality when I wanted to aggregate my commits and found this to be a use case for StaticXSLT
* feeds are fetched via `curl`
** this will be a local request in production
** the messages are still cleaned using `sed` and transformed into _XHTML_ using `peg-markdown`
* instead of reading a customized Git log output CommitLog now simply transforms CGit ATOM feeds using identity templates
** this requires ATOM namespace removal which is implemented in `utility/remove_namespace.xsl`

* moved repository feeds to `repository` directory
* replaced paginated XHTML timeline target with and ATOM feed
** removed `paginated_timeline.xsl` transformation including the `04_meta` level
** removed CSS and master stylesheets
* added title, url and commit count meta data to `meta.xml`
* expanded ATOM feed URLs
* timeline feed has an upper limit of `commit_count` items
** repository feeds have the same limit because only `commit_count` commits of every repository are fetched in the first place

* another application based on StaticXSLT
* git is instructed to export XML through a special log format definition
* the commit messages are processed as Markdown
* currently implemented result views are commits-by-repo and a paginated timeline of all commits in every repository
* repositories to be read have to be defined in the `repositories.xml` file in the `00_content` level