scan_package should not fail if there is an error in the scan

[pombredanne]

When scanning https://codeload.github.com/ElinamLLC/SharpVectors/tar.gz/refs/tags/v1.8.2 with "scan_package" the pipeline fails with:

{'codebase/SharpVectors-1.8.2/Source/SharpVectorModel/Compressions/Brotli/Dictionary.cs': ['ERROR: for scanner: copyrights:\nERROR: Processing interrupted: timeout after 120 seconds.']}

Traceback:
  File "/opt/scancodeio/scanpipe/pipelines/__init__.py", line 121, in execute
    step(self)
  File "/opt/scancodeio/scanpipe/pipelines/scan_package.py", line 109, in run_scancode
    raise ScancodeError(errors)

This should not fail (since this is a recoverable error) but should instead be loaded with the errors as error messages and the scan as output.

aboutcode-org/scancode-toolkit#3614

[tdruez]

Improved in #1019