Add delayed parsing function to use in flex-tasks parsing

[patritzenfeld]

To execute only once and receive either an error message or the result, then use this value for both syntax and semantics checks.

[patritzenfeld]

Changed that .md file because the Xrefchecker complained about a redirecting link and failed.

[jvoigtlaender]

Ich denke, die Funktion sollte im Sinne von fmidue/flex-tasks#72 (comment) vereinfacht werden. (inlining von delayed etc.)

Und dann stellt sich noch die Frage, ob die Funktion hier in logic-tasks bleiben sollte oder in fmidue/flex-tasks#77 integriert. Wenn Delayed nicht mehr im Spiel ist, könnte das ja durchaus als eine Funktion angesehen werden, die allgemein in flex-tasks Konfigurationen nützlich ist, nicht auf Logik-Aufgaben beschränkt.

[jvoigtlaender]

I don't see why this function has Delayed in its name, given that none of its arguments does actually involve any Delayed business.

And also its implementation has no reason to use anything like delayed or parseDelayed etc. It can simply be written like this:

newFunctionName p messaging fallBackParser =
  parseWithOrReport
    (parse (fully p) "(answer string)")
    (messaging . either Just (const Nothing) . parse (fully fallBackParser) "(answer string)")

I'm actually uncertain whether the fully calls are really necessary, and whethere "" instead of "(answer string)" would make sense.

But the main point is: This function isn't about Delayed parsing at all. And also it does not really belong in the logic-tasks repo. It isn't used in logic-tasks and it also isn't specifically about parsing formulas or other logic concepts. It is just a certain way of doing parsing with a fallback for better error messages.

[jvoigtlaender]

(After this function is moved to flex-tasks, it is possible that parseDelayedWithAndThen below can be expressed in terms of it. That is, importing newFunctionName from flex-tasks instead of importing parseWithOrReport from there. But that is really a different aspect.)

[jvoigtlaender]

This pull request has now essentially become a closer for #217 instead of what it started out as.

[jvoigtlaender]

(Might be better to close it and make a clean new pull request addressing #217.)