Import content from PDFs and automatic chunking support

[monatis]

letsearch should support configurable automatic chunking when importing unstructured content from PDFs and other docs.
The user should be able to choose whether to store chunk vectors with a single common key.