Sometimes, in an ordinary perfect cadence the leading tone doesnt resolve but moves to ^5 if it is in an inner voice. So why does it have to resolve to ^1 in a deceptive cadence or V-VI progression?
First of all a leading tone should resolve if possible. A leading tone falling into the fifth is only really considered an option for an inner voice to ensure that the final chords is a full chord. And this is only really relevant if the fifth of the dominant is in the top voice, which means that the root of the dominant has to fall into the third of the tonic, and thus the fifth would be missing.
Now for a deceptive cadence first see that the leading tone (if not resolving) would need to fall into the root of the VI, which is already played by the bass. So there is no actual reason for this. Also if we have V7-VI this would then imply a tritone moving in parallel into a fourth (not very elegant) or into a fifth (to be avoided).
In a V-VI the root of the V (or in case of a V7 the seventh) needs to fall into the fifth of the VI, and the fifth of the V needs to fall into the third of the VI. So if the leading tone resolves it as to rise into the third of the VI too. So you will always have to double the third (which is the root of the tonic). While you could probably use a falling leading tone to circumvent this, one has to see that this nescessary doubling is in fact an important tonal characteristic of the desceptive cadence, as since you have the root of the tonic twice it has a stronger cadence like character than you’d get if you had the full triad on the VI.
The main issue is that of the augmented second in minor. If the leading tone of the V chord moves down to a member of the VI chord, the nearest possible chord tone is that lowered scale-degree 6, an augmented second away from the leading tone.
As such, this leading tone moves up to double the third of the submediant chord.
In major, this augmented second isn't a problem—the leading tone could just move down a major second to scale-degree 6. But in my experience, composers often automatically resolve this leading tone up anyway.