Translating FrameMaker documents with insets

Hi FrameMaker experts,

I've searched in the forum and also read What is the best method for language translations?, but I cannot find the answer I'm looking for. Please excuse, I'm not very familiar with FrameMaker. 

Environment: Unstructured FrameMaker 15, FrameScript is available

We try to enhance the number of insets to reuse more content and get smaller snippets. So far, all modules are stored in FrameMaker files, which are normally not being sent for translation within a project, only if the technical writer manually picks those files. All insets are redirected with a FrameScript to "empty module files" before translation and then redirected to the orginal module files when the translated MIF file is converted to a FrameMaker file again. This works and gives the technical writer more control of what is being translated. But the problem is, it is too complex for an environment with many insets.

FrameMaker seems to copy "something" from the module file (referenced by the inset) to the MIF, but "not really" the whole file content. I mean, I cannot unpack the MIF and get all files back, which were referenced with insets, only the main FrameMaker file.

 <Para 
  <Unique 998513>
  <PgfTag `Haupttext'>
  <ParaLine 
   <TextRectID 19>
    <TextInset 
    <Unique 998510>
    <TiSrcFile `<c\>my-module-file.fm'>
    <TiLastUpdate  1596632557 428000>
    <TiAutoUpdate Yes>
    <TextInsetLocked No>
     <TiFlow 
     <TiMainFlow Yes>
     <TiFormatting TiEnclosing>
     <TiFormatRemovePageBreaks No>
     <TiFormatRemoveOverrides No>
    > # end of TiFlow
   > # end of TextInset
   <String `this is the content of my-module-file.fm'>
  > # end of ParaLine
 > # end of Para

So, it seems I have to manually select the module files and copy them to the set of files to be translated. The FrameMaker packaging function seems to pull/pack all the inset files, but it seems this packaging function is not meant to be used for translation.

So, here is my question: If I have a complex file structure, many documents and hundreds of insets in a document pointing to many different files in many different directories, how do I translate such a monster? Do I have to handpick all files? Or would it be easier to just break all resolve/break all insets, so copy the referenced content to the FrameMaker file and send this file without any insets to the translation agency? Is there are more efficient approach without loosing insets in my translations (mirroring the file structure of the source language files)?

Thank you very much for sharing your expertise,

Stefan