Fix approval in manual operations

[guillaume-duong-bib]

Fixes mitre/caldera#2887.
Fixes and closes mitre/caldera#2962

Reasoning for the changes:

• deletion of line 31:
  • Just by thinking about it, this doesn't make much sense to me, since it basically means "if the link is not finished (does not have a finish date), it is not editable". This also goes against the comment that mentions that links in paused or queued can also be editable.
  • Secondly, this contradicts the last check, because we end up in this situation:
    • (currentLink.finish = False) => (isEditable = False) (from line 31)
    • (currentLink.finish = True) => (isEditable = False) (from line 36), assuming that (currentLink.finish.length > 0) <=> (currentLink.finish). I assume this because I could only find any assignment to link.finish with the value of get_current_timestamp(). Since the initial value of finish is None, this would mean finish either has a value of None or a strictly not empty string. So right-to-left is True, with left-to-right being trivial.
    • Assuming both propositions are true, we always have isEditable = False (and thus the Approval is never possible).
• modification of line 36:
  • By reformulating !(currentLink.finish.length > 0), we are checking that currentLink.finish is an empty string. But as explained before, this situation does not happen. If finish is not set, it has a value of None/Null and thus .length fails.

The only issue I could imagine with this fix is in the event of finish being able to reference an empty string for specific situations, but I did not find any trace of that ever happening.

Here's what a manual operation looks like with this change:

NB: as per mitre/caldera#2968, using this fix will in most cases trigger another error, but I chose to leave it separate as it's a different root cause with a different fix.

[elegantmoose]

on my queue

[elegantmoose]

Your logic tracks. Im guessing it was a copy/paste error, using the line above and then forgot to remove the "!". Removing the line all together seems cleaner too given the final compound logic evaluation.

[elegantmoose]

Requires #49 as well to work.

[elegantmoose]

Thank you much @guillaume-duong-bib ! 👏