Changed handling of blank environment, fixed section handling on command line, cleaned up testing, added release changes

* Changed behavior if environment name is set but empty: now uses default instead of looking for '' environment
* Fixed command line handling: different variables in the same section are now all recognized (see IssueID #5501). The test cases are updated accordingly
* Re-organized test cases to group command line handling in its own class
* Added empty [namelists] section to test config, simplifying test writing
* Added CHANGES.txt for upcoming release 0.3.0, updated setup to 0.3.0dev

* Added traversal (walk) function to parse keys and values for verbatim target variables, i.e. variables of the generated script language
* Added environment functions to temporarily reset the string used for new lines. Needed to work around the missing 'wrapstring' keyword in Jinja 2.6
* Added context function to allow querying context variables by name within templates. Needed to get default values for target variables.

Added configurable variable format to namelist values. Variables $${x} are now recognized as script native variables. Defining .var_format allows different script syntax, e.g. XML style (.var_format = <%s>)

Fixed override behaviour for global job config: now specialized job config will always override global job config

* Added new control variable 'EXP_OPTIONS', taking a list of config names. Options are taken from the new 'standard_options' subdirectory
* Added respective test and simplified test module

* Added setting of sectioned variables via command line, using '.' as section path separator and '..' for verbatim periods
* Added automated testing using Python's unittest library
* Disambiguated warning due to missing experiment configuration; now mentions 'experiment'. 

* Added python methods list.split() and filter() to template environment for advanced template macros (file lists)
* Changed job script template search to allow local <EXP_ID>.<JOB.id>.tmpl override standard experiment templates.

* Changed to use directories for backup, leaving file names unchanged
* Added README file to backup

* Logging
** Fixed bailout routine to allow message format arguments
** Explicitly added default warning level for ease of debugging
* Added verbose mode to getexp, showing all top level variable settings
* Fixed template loading to allow out of directory generation (-p, MKEXP_PATH)

* r3545: Changed to parse namelist section comments for example key assignments.

Unified handling of remove keywords in job configuration, added remove keyword for namelist variables

* r3359: Script generation: improved documentation (README, user guide), fixed time limits, number of tasks override
* r3358: Downgraded automake use to 1.13.1 installed as module in squeeze.
* r3357: J.S.Rast: Added missing profile viewgraphs.

* r3341: Changed comments in namelist to deal with inline comments and empty lines

* Moved mvstream settings to default configuration
* Adapted examples accordingly and updated README and docs
* Changed blizzard setup to use /scratch for working, read account from .acct

* Added missing file to r3147

* Changed to use argparse library for command line handling
* Added option to set search path for standard configuration (--path)
* Added use of config file comments in namelist files

* Added support for search path to getexp (amends r3166)
* Added support for keyword substitution to default environment (amends r3061)

Changed script generator to allow a path for config files and templates (MKEXP_PATH); directories are separated by colon.

* Added optional template for namelists (namcouple)
* Added info line for generated files (mkexp_input variable)
* Added model directory to 'getexp' output

* Added namelist defaults for afterburner select lists
* Change mkexp to allow variables in top-level namelist definition

* Merged r3149: Fixed cleaning of packaged CDI library. from trunk/echam-dev
* Added removal of config.inc, stamp files
* Added mkexp sub directory to distclean chain

* Changed job extension to also inherit job settings (used to inherit script only)
* Changed job extension to allow namelist overwriting per job
* Namelist overwriting requires copying of the configuration data; this should be revisited

Changed script generation to allow suppressing of default jobs by 'remove' keyword in 'jobs' section

Changed script generation to allow easier namelist setup (helper variables in namelists, evaluation of lists as strings; see EaSyMS13 IssueID #3621)

* Changed to support list valued expressions
* Changed to support expressions in lists

* Removed unused scripts
* Fixed module loading
* Anonymized default settings