init: explicitly send err output to sys.stderr in python

Update python initialization script to explicitly send error output
to the sys.stderr channel (instead of parent process stderr channel) to
be able to catch this content.

init/python.py.in

@@ -1,4 +1,5 @@
-import os, re, subprocess
+from __future__ import print_function
+import os, sys, re, subprocess
 
 # define modules runtime quarantine configuration
 @setquarvars@os.environ['MODULES_RUN_QUARANTINE'] = '@RUN_QUARANTINE@'
@@ -21,7 +22,13 @@ if 'MODULES_RUN_QUARANTINE' in os.environ:
 
 # define module command and surrounding initial environment (default value
 # for MODULESHOME, MODULEPATH, LOADEDMODULES and parse of init config files)
-exec(subprocess.Popen(['@TCLSH@', '@libexecdir@/modulecmd.tcl', 'python', 'autoinit'], stdout=subprocess.PIPE, env=_mlre).communicate()[0])
+out, err = subprocess.Popen(['@TCLSH@', '@libexecdir@/modulecmd.tcl', 'python', 'autoinit'], stdout=subprocess.PIPE, stderr=subprocess.PIPE, env=_mlre).communicate()
 
-# clean temp variable used to setup quarantine
+exec(out)
+if err.decode():
+   print(err.decode(), end='', file=sys.stderr)
+
+# clean temp variables
 del _mlre
+del out
+del err