ocrd-tool schema: Describe "description" and the semantics of in/output_file_grp

[kba]

The meta-documenations is a bit meagre at the moment, adding (better) descriptions to all fields will help implementers write more expressive ocrd-tool.json.

https://github.com/OCR-D/spec/pull/121#discussion_r309177809 ff.

[bertsky]

Regarding input_file_grp and output_file_grp: the latter has already been made non-required. However, that's not because we do not see much use in specifying fileGrp names statically anymore. But rather because we found that not all processors may actually have an output fileGrp.

But if the current semantics (Input fileGrp@USE this tool expects by default / Output fileGrp@USE this tool produces by default) is correct, then why make these required at all? (It just adds clutter to the ocrd-tool.json definitions without any actual value IMO.)