Handling nested folder structure and directly overwriting originals

[jotelha]

For me, a few modifications were necessary to

• have it properly overwrite originals if in- and output PDF are the same:
  previously, the content was lost in such a case. Now, this is circumvented via a temporary file.
• preserve a possible nested multilevel folder structure containing the PDF:
  my PDF were not within the Mendeley data folder, but somewhere else, within author name subdirectories.
• handle exotic UTF-8 file names properly:
  it did not read or write i.e. Chinese PDF names on my system. Using
  .encode(sys.getfilesystemencoding()) wherever a file is accessed or the name displayed works fine.

Also, the README now references several Mendeley -> Zotero migration guides, introducing the approach that worked for me.

[cycomanic]

thanks merged