bake: Better error handling for the processing pipeline. Pipeline jobs now keep track of whether they've seen any errors. This is aggregated into an overall "success" flag for the processing record. Also, jobs keep going as long as there's no critical (i.e. internal) failure happening. Errors raised by processors are also better tracked: the actual processor that failed, along with the input file, are tracks in the processing record. The `bake` command returns a failure exit code if processing saw any error.

--- a/piecrust/processing/records.py	Sat Jan 31 16:28:16 2015 -0800
+++ b/piecrust/processing/records.py	Sat Jan 31 17:08:02 2015 -0800
@@ -9,6 +9,7 @@
         super(ProcessorPipelineRecord, self).__init__()
         self.out_dir = None
         self.process_time = None
+        self.success = False
 
     def hasOverrideEntry(self, rel_path):
         return self.findEntry(rel_path) is not None