fixed some bugs in api key auth, added access by api key into rss/atom feeds in global journal
and repositories feeds

Fixed Windows installation based on work of Mantis406 fork: "Replace py-bcrypt to make Windows installation easier"
RhodeCode will now detect the platform and use sha256 hashes when used on Windows.

Major rewrite of auth objects. Moved parts of filling user data into user model.
Rewrote AuthUser adding access by api key.

Added api_key into user, api key get's generated again after password change
updated ini files

#109, added manual pull of changes for repositories that have remote location filled in.
This also logs this action in journal

#109, added optional clone uri when creating repo.
Cleaned up repos controller moved common code into __load_data

Added branch filter to repo pager
fixed stop condition for building statistics

fixed problem with rawfile content_disposition attachment,
webob assumes that file is encoded with iso, and we use utf8 in rhodecode

replaced all decode('utf-8') instances with .decode('utf-8','replace') for more error prof setup,
this way rhodecode could handle displaying non utf8 encoded file paths. This is still an invalid path, but this way we could at least show those paths without errors

fixed unicode problems with file paths.
added feed arg to journal action mapper

implemented public journal for anonymous users, admin can control which repositories
are visible in such journal in admin panel

fixed Session problems in model class functions
added dummy mapper extension for repository

moved statistics parse_limit into .ini files
decreased celery max task per child to 1 for better memory usage
stats task take usage of new slice option of repository

Fixes for release 1.1.4
Docs updates

Moved BaseModel into base class for declarative base. Added some handy methods into
BaseModel like get() and query()

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

fixes fixes fixes ! optimized queries on journal
added some logging, fixed found bugs due to previous refactorings

extended admin rescan to show what repositories was added and what removed
fixed found bugs, added expunge to cached getter since it introduce some more problems.
Added some relationship to properly clean related objects from database.

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

Major refactoring, removed when possible calls to app globals.
Refactored models to fetch the paths needed for scans from database directly
small fixes in base