Add LangPro predicted entailment label

[XanderVertegaal]

The user currently has to inspect the LangPro-generated tableaus to see what LangPro's judgement of a problem is (contradiction, entailment, neutral or conflict). It would help if this could be displayed somewhere on the page. It should not overwrite the existing entailment label that is part of the dataset.

The entailment label can be inferred from the tableau output:

Entailment tree closed	Contradiction tree closed	Resulting judgement
✖️	✔️	Contradiction
✔️	✖️	Entailment
✔️	✔️	Conflict
✖️	✖️	Neutral

This predicted entailment label should only be saved to the DB once 'Save problem' is pressed.

[jgonggrijp]

Amendment: the saving of the prediction should also only be saved if the current tableaux are based on the set of KB items that are included in the problem save.

[XanderVertegaal]

Update:

Entailment tree	Contradiction tree	Resulting judgement
open or failed	closed	Contradiction
closed	open or failed	Entailment
failed	failed	Unknown
open or failed	open or failed	Neutral
closed	closed	Conflict

We can call this "LangPro prediction" and distinguish it from the entailment label that is part of the datasets (SICK, FraCaS, SNLI), which should be renamed to "Ground-truth label". There should be a filter for this LangPro prediction (similar to the one we have for the ground-truth label. The prediction field should be nullable, for problems that have not been proved yet, and the filter should allow for searching for problems without a prediction.