This is not an a totally absurd proposal if dosage and time-related issues weren't a factor, but it is rife with practical difficulties otherwise.
In schizophrenia patients, what you're talking about has been widely studied, under terms like "antipsychotic-evoked dopamine supersensitivity", "supersensitivity psychosis" (SP) etc. The latter is basically "a limbic equivalent of TD".
The problem is that TD and SP are pretty correlated, in terms of the dosage required to obtain them, and require "overdosing" the (schizophrenia) patient. Typically, in order to obtain the anti-psychotic effect, 65%-80% peak striatal D2 occupancy is needed. TD/SP most commonly occur at greater than 80% occupancy.
Using 1st gen anti-psychotics (e.g. haloperidol) for this purpose in therapeutic is probably the worst idea, because achieving the desired super-sensitivity to striatal dopamine would mean giving the patient TD (tardive dyskinesia) with pretty high probability. TD is generally considered the worst side-effect of anti-psychotics (see previous link). You could say the cure (TD) is worse than the disease you'd want to treat (ADHD). On other hand, it may a little more plausible to do this with atypical (2nd gen) antipsychotics.
For instance, quetiapine  and clozapine [33,40] can induce SP without producing overt or detectable drug-induced movement disorders.
But atypical anti-psychotics aren't totally free of the risk of TD, they just reduce it.
Even disregarding side-effects (which AliceD's answer covers aplenty), which would actually be exacerbated by the high dosage needed, the problem is that the SP phenomenon (which would be the basis of propose cure for ADHD) is variable in time:
The persistence of the dopamine supersensitivity syndrome depends on the duration of the preceding blockade  and on the specific antipsychotics used (fluphenazine, perphenazine, clozapine, and quetiapine) [1,41].
So it would not be a cure for ADHD except possibly in small number of patients for whom the long-term effect comes out "just right". There's also the obvious risk of "overshooting" and inducing actual likelihood of psychosis in someone (with ADHD) who didn't have such a risk.
In general, giving a patient a drug and adjusting the dosage until the desired effect is observed is simpler than giving a drug hoping for permanent effect that only stabilizes some time after not taking the drug.
On top of these, there's the issue that the exact mechanism of ADHD isn't absolutely sure. And that basically all psychotropics act on more than one neurotransmitter, so the don't totally cancel each other out. For these reasons, there are even publications discussing co-medicating some/few patients with anti-psychotics and stimulants, even though these classes of drugs partially cancel each other out.
Now very interestingly, there is a 2014 paper (alas with only 6 citations) that does propose using very low doses of haloperidol as an adjuvant of L-Dopa in Parkisons precisely for this purpose of causing dopamine supersensitivity (and not for the common purpose of treating co-morbid psychosis that some Parkinson patients exhibit).
The trick that makes very low doses feasible for this purpose is that antipsychotics cause the D2 receptors to switch to a D2High (high-affinity) state. Apparently this switching happens even a low dosage of antipsychotics.
Unlike in your proposal though, although the effect apparently lasts for a while after haloperidol discontinuation, it begins on it.
The authors of this paper also emphasize a more substantial decline in "freezing" (a Parkinson symptom) that may coincide with the peak (and delayed) effect of D2High caused by haloperidol, but frankly the data is pretty weak in that regard and one can easily level "p-hacking" at this particular finding:
If it needs saying, even if this "trick" were to work better (e.g. with other doses), since Parkinson is neurodegenerative, there's only so much tweaking at the margin of the remaining neurons' sensitivity you can hope to do, before the loss of more/enough neurons overtakes that extra sensitivity.
Also, there actually many ways to trigger D2High states, stimulants themselves (e.g. amphetamine) do that.