First step in two-part process to rename directories to kallithea.
This first step is to change all references in the files where they refer
to the old directory name.

Imported some of the GPLv3'd changes from RhodeCode v2.2.5.

This imports changes between changesets 21af6c4eab3d and 6177597791c2 in
RhodeCode's original repository, including only changes to Python files and HTML.

RhodeCode clearly licensed its changes to these files under GPLv3
in their /LICENSE file, which states the following:
The Python code and integrated HTML are licensed under the GPLv3 license.

(See:
https://code.rhodecode.com/rhodecode/files/v2.2.5/LICENSE
or
http://web.archive.org/web/20140512193334/https://code.rhodecode.com/rhodecode/files/f3b123159901f15426d18e3dc395e8369f70ebe0/LICENSE
for an online copy of that LICENSE file)

Conservancy reviewed these changes and confirmed that they can be licensed as
a whole to the Kallithea project under GPLv3-only.

While some of the contents committed herein are clearly licensed
GPLv3-or-later, on the whole we must assume the are GPLv3-only, since the
statement above from RhodeCode indicates that they intend GPLv3-only as their
license, per GPLv3§14 and other relevant sections of GPLv3.

invalidation: don't create CacheInvalidation records on startup

Creating the records early gave an advantage before lightweight was introduced.
With lightweight it is no longer necessary.

The records will be created on demand anyway and there is no reason to create and
maintain them before they are used.

invalidation: merge .invalidate and .set_valid as .test_and_set_valid

This simplifies the internal API and avoids leaking CacheInvalidation instances
and cache keys.

Removed shortlog aka lightweight changelog.
Current version of changelog is fast and condensed.
There's no sense to keep changelog duplicity.

auth decorators are not used anymore on __before__
- this will allow to create a whitelist of API enabled controller functions
- moved around functions with _ to beginning of the files to make it cleaner

Repo size - show just the size without duplicating text
bring back numeric revision into changelog
Fixed tests

Addding context bar to more repo related pages.
Fixing the base template to work on non-repo related pages.

added recursion limit for stats gathering, sometimes it did >1000 loops which lead to python throwing max recursion depth exceeded error.
ref #642

Added uri encoding to the username so that usernames with '@' in them (such as email address/users from LDAP) will
not have to replace the first '@' with %40 in order to clone

Bumped mercurial version to 2.3
- fixed issues with compare of two remote repos at particular revisions
- added tests for cases when mercurial update broke code and tests didn't catch this
- removed obsolete EmptyChangeset from the rhodecode code

Readme renderer now uses landing_rev parameter to render the readme based on
given revision, rather than tip

fixed readme rendering bug.
when there existed a directory named like dicovery file for readme, it crashed trying to read it's content. We just skip them now

utils/conf
- created temporary utils2
- made config.conf for storing some configurations
- fixed some dependency import problems
- code cleanup
- rc-extensions now properly work for celery

fixes issue #271
- OrderedDict serialization sometimes failed due to how odict is implemented. Dropped usage of it for a simple tuple

backported fixes for statistics from beta branch, try to fix issue #271
- version bump

- fixed issue with missing commits on some repos commands
- fixed issue with parsing commits for stats when there were only two commits

configurable clone url
- posibility to specify ssh:// manually as alternative clone url.

Cached readme generation
- cleaned up cache system

fixed issues with python2.5
added compat module to rhodecode to have one point of fetching backward incompatible libs.

Fixes issue #201
Fixed bug when repository was deleted on filesystem, and not from rhodecode

fixed issue #165 trending source files are now stored in cache as ext only, and translated to description only when displaying, so future changes of mappings will take affect right away.
Also changes the way how map is generating, ie. autogenerate from pygments lexers, and than complete with user defined mappings.

another major codes rewrite:
- created BaseRepo controller for all repo specific controller, and added common data propagation
- removed obosete codes, and made optimizations, removed to often calls to RepoModel
- fixed found bugs in files controller that generated unhandled 500 errors
- cache issues
- removed repo_branches global template values
- journal fixes
- main repo list is fully dict now it's less resource heavy that way

another major code rafactor, reimplemented (almost from scratch)
the way caching works, Should be solid rock for now. Some code optymizations on scmModel.get() to make it don't load unneded things. Changed db cache to file that should also reduce memory size

Rewrite simehg for enabling cloning with raw url for anonymous access + some optimizations for making less queries when authenticating users.
added debug to test hg operations

extended trending languages to more entries, implemented new faster and "fancy"
method of extensions recognition.
Fixed small bug in indexer when data dir was empty

Refactor codes for scm model
Some test updates, added test for admin user controller

updated setup for all newest versions
EmptyChangeset can take changeset as creation param.

removed shortlog from main menu, as duplicated functionality of changelog,
moved shortlog as ajax pagin in summary

Code refactoring,models renames
cleaned up sqlalchemy sessions,
added cache support to most queries in models
fixed test.ini file

#37 fixed json imports for python2.5
removed cache manager settings out of meta to ini files
docs update.