This includes changing a lot of routes and parameter names to be more
consistent. We try to remain backwards compatible as much as possible
by providing redirects from old routes and parameter names.

Now that we ported the database to SQLAlchemy, we remove the obsolete Db
module and all references to it.

The Redis client automatically falls back to a mock Redis server if no
Redis server is configured. Therefore, a Redis server is not needed to
run Mutalyzer. You'll just not get any aggregate stat counts over
different runs.

This introduces a proper notion of genome assemblies. Transcript
mappings for alle genome assemblies are in the same database, which
is better for maintenance. Updating transcript mappings is also
simplified a lot, especially from NCBI mapview files where we now
require a preprocessing sort on the input file.

Overall, this port touches a lot of Mutalyzer code, so beware.

See Trac ticket #148

This involves making the SQLAlchemy session reconfigurable at run-time,
which is done automatically on updating the Mutalyzer configuration using
configuration update callbacks.

Port the entire batch job infrastructure, including scheduler, to use
the SQLAlchemy ORM instead of the old Db module.

Warnings were logged with message level 3, which is for errors, and
therefore interpreted as errors (ignoring the batch entry result).

Remove the dependency on configobj and have default values for all
configuration settings. User settings are defined in a Python module
pointed to by the MUTALYZER_SETTINGS environment variable.

We also clean up many configuration settings and remove some that
are no longer used.

We no longer look for /etc/mutalyzer/config and ~/.config/mutalyzer/config
since it is too low level and inflexible. The user should now specify the
location of the configuration file in the MUTALYZER_SETTINGS environment
variable (or the file mutalyzer.conf in the current directory is tried).

We should leave process control to others, for example supervisord which
can automatically restart crashed processes.

Do not try to implement a full UNIX daemon suitable for init.d with the
batch daemon, but rather keep it simple and depend on something like
supervisord to implement process control.

We now also correctly handle SIGINT (Ctrl+C) and SIGTERM signals.