es-dedupe

Completely rewritten implementation.

• Support removing duplicates by multiple fields
• Using sliding window for timestamp ordered documents

• Fixed Docker image

• Reorganized ENV variables

• refactored code, should work with ES 5.x
• fixed StringIO import for python2 and python3 (imports)
• added better logging (logme)
• added more log messages
• revamped verbose/debug output (debug=more output)
• made sure the indexname is always called the same all over the script (idxname)
• removed date-counting completely (inc_day, msg_using)
• implemented getting the full indexlist instead, including excludes via regexp (idxlist_uri, fetch_indexlist, -I)
• implemented function calls failing (rc = -1) instead of interpreting rc = 0 as an error (which it is not when an index-query returns 0 entries on a readable ES index)
• in case an index has blocks->write=true set, we temporarily set it to false, do our dupe deleting and set it back to what it was before our changing it (settings_uri, allsettings_uri, fetch_allsettings, set_index_writable)