Simplify reduction operation implementations and error handling

(Targeting feat/compress-keys for now, as it requires significant refactoring to work for trunk right now)

Simplifies the implementation of reduction operation by moving the boilerplate iteration over applicable sources and/or keys to some generic machinery, as well as adding some custom exception types to propagate fatal error conditions from the ReduceWriter initializer back to the CLI and exit gracefully without an unchecked exception.

src/exdf/cli/reduce.py

@@ -12,7 +12,7 @@ import sys
 from pkg_resources import iter_entry_points
 from extra_data import RunDirectory, open_run
 
-from ..data_reduction.writer import ReduceWriter
+from ..data_reduction.red_writer import ReduceWriter, ReduceInitError
 
 
 def _parse_args(argv):
@@ -212,9 +212,12 @@ def main(argv=None):
     if args.to_recipe:
         _to_recipe(args.to_recipe, methods, inp_data, argv)
 
-    writer = ReduceWriter(
-        inp_data, methods,
-        args.output_scope, args.output_sequence_len, args.output_version)
+    try:
+        writer = ReduceWriter(inp_data, methods, args.output_scope,
+                              args.output_sequence_len, args.output_version)
+    except ReduceInitError:
+        log.critical('Failed to initialize reduction writer')
+        return
 
     if args.output_scope == 'none':
         log.info('Not writing out any data files')