- introduced a timer so data are not instantely reloaded
 - corpus list is made more compact
 - filters on metadata seem to be working properly
 - some fields are explicitely rejected for metadata filtering