I think an appropriate concept here is "contra proferentem". Essentially, where there is ambiguity this counts against the author, not the reader.
If you want a unique, "correct" answer, then you must construct your problems such that they each have only one valid answer. This is a significant component of problem setting. There is very little in the world of puzzling as infuriating as wrestling with a problem, finding a perfectly valid solution, and then having it labelled "incorrect".
Let's say I set the crossword clue "disappointed on big day". You might very well get the answer I intend from that. I would go so far as to say that you would probably get the answer I intend. But you could also get a dozen others that I don't intend. If I write "Short detective returns to Jack's partner left at altar, we hear", there are far fewer possibilities*.
Where a problem has multiple solutions, I would suggest that, as a minimum, answers should be rewarded for showing any of the qualities required to obtain the intended answer. Beyond that, I see nothing wrong with rewarding creativity and lateral thinking; many riddles are designed in this way, and if you want a different kind of answer you should construct your problems accordingly.
In general, good problems have solutions that are difficult to guess, but easy to accept; ideally, they are provably correct. For problems with a supposed "correct" answer, solutions should be unique.
(*A thousand honorary upvotes to anyone who gets a valid answer I didn't intend... Edit: I've added an explanation of the intended answer to use as a springboard. Note how all parts of the clue are accounted for in the solution.)
[Short detective] [returns to] [Jack's partner] [left at altar], [we hear]
"Short" could indicate an abbreviation:
[Short detective] ---> [Det.]
Popular nursery rhyme: "Jack and Jill went up the hill..."
[Jack's partner] ---> [Jill]
[returns to] implies a part of the clue running backwards and towards another part of the clue:
[Det.] [returns to] [Jill] ---> [Jill] [ted]
Note: not [ted] [Jill] because the letters "d", "e", "t" run away from [Jill]
[we hear] indicates a sound-alike; if someone says "Jill-ted", [we hear] "jilted"
[Jill] [ted] ---> [jilted]
The above parts of the clue form the indication or wordplay. What remains, [left at altar], is the definition. Both parts point to the same solution, "jilted".