Remove most of task evaluation code from Task class, rely on
Task::DSL.evaluate to get an evaluated task.
* Task: remove #evaluate, change constructor prototype to accept actions
instead of a block, implement .evaluate(name, env &block);
* Implement Task::DSL.evaluate method;
* Recipe::DSL: remove tasks evaluation from#evaluate, modify #task to
use Task.evaluate to return the new task to be rigstered.
* Throw symbols instead of raising exceptions to test code evaluation,
this will reduce "bad" usages of exceptions, symbols seems a little
better for now;
* Add a specific exception to be used in specs for testing behaviour
when an exception is raised from a recipe.
* Report invalid action usages from tasks;
* Implement backtrace cleaning in CLI;
* Extract error class declarations in a new errors file;
* Replace raise with fail keyword in task DSL class.
