SCTK reports declared_holder incorrectly when a LICENSE file is a license text copyrighted by the Free Software Foundation

[DennisClark]

A recent scan of MinerU-magic_pdf-0.6.1-released.tar.gz downloaded from https://github.com/opendatalab/MinerU/archive/refs/tags/magic_pdf-0.6.1-released.tar.gz returns misleading copyright information. The project is clearly licensed under agpl-3.0 as the full text of that license is provided in a LICENSE.md file at the top level of the project; however, the project authors have not provided a Copyright statement for the project as a whole although one might assume that OpenDataLab, the owner of the GitHub project, is also the copyright holder, but that is not declared or stated. SCTK incorrectly returns "Free Software Foundation" as the declared_holder, even though the FSF is only the copyright holder of the license text itself, not what the license is applied to. This leads to a further problem in that the declared_copyrights field is then set to true, leading to a misleading license_clarity_score of 100; instead, it should be set to false, and the resulting license_clarity_score should be 90. Details attached.

MinerU-magic_pdf-0.6.1-released.tar.gz_scan.zip

[pombredanne]

https://github.com/aboutcode-org/scancode-toolkit/blob/develop/src/cluecode/plugin_filter_clues.py is already there and should be working there.

[pombredanne]

We need to be doing a quick pass on ignore clues in license rules so we can determine if some should NOT be ignored and we should make this the standard.