Rework scan-archive multiprocessing scan

This is an important rework of the code in s01scan_archive.py to make it
more readable, easier to maintain, and more efficient while scanning
files and directories.  (This was first initiated to correct some
multiprocessing-related bug/odd behaviour on CentOS 6.5).

It uses multiprocessing.Pool objects to distribute the scan of the
archive directories among multiple child processes, if the -t/--threads
option was used to indicate the number of concurrent processes to use.

The logging system was also slightly changed in s01scan_archive.py and
scripts/msnoise.py.  A logger instance is now setup through the
api.get_logger() function (and a new instance is created in child
processes to add the process id in the log record format).

msnoise/__init__.py

@@ -11,6 +11,8 @@
 class MSNoiseError(Exception):
     pass
 
-
 class DBConfigNotFoundError(MSNoiseError):
     pass
+
+class FatalError(MSNoiseError):
+    pass

msnoise/api.py

@@ -11,6 +11,7 @@
     import pickle as cPickle
 import math
 import pkg_resources
+import sys
 
 from sqlalchemy import create_engine, func
 from sqlalchemy.orm import sessionmaker
@@ -33,6 +34,29 @@
 from .msnoise_table_def import Filter, Job, Station, Config, DataAvailability
 
 
+def get_logger(name, loglevel=None, with_pid=False):
+    """
+    Returns the current configured logger or configure a new one.
+    """
+    if with_pid:
+        log_fmt='%(asctime)s msnoise [pid %(process)d] '\
+                '[%(levelname)s] %(message)s'
+    else:
+        log_fmt='%(asctime)s msnoise [%(levelname)s] %(message)s'
+    logger = logging.getLogger(name)
+    # Remove any inherited StreamHandler to avoid duplicate lines
+    for h in logger.handlers:
+        if isinstance(h, logging.StreamHandler):
+            logger.removeHandler(h)
+    handler = logging.StreamHandler(sys.stderr)
+    handler.setFormatter(
+            logging.Formatter(fmt=log_fmt, datefmt='%Y-%m-%d %H:%M:%S'))
+    logger.addHandler(handler)
+    logger.setLevel(loglevel)
+    logger.propagate = False
+    return logger
+
+
 def get_tech():
     """Returns the current DB technology used (reads from the db.ini file)