I conclude these are likely quite individualized effects -
those of one's own metabolism (?) from limited hearsay, all along.
While a diet is not a 'fast', there are similarities. Can only report from direct experience:
Some years back (for no very compelling reason, at the time - other than just general 'cleaning out') I embarked on a very-simple fast regimen. Partly I was curious, too - as the person with recipe had used this method, while suffering some local hepatitis strain + jaundice in India (!) -- to cure that situation, and also to evade being in hospital there, on IV and many etcs. It worked for her, and in her hotel room - no pharmchem. (Again - no general-panacea implied.)
Regime was: real maple syrup (not the ersatz corn- toxified substitutes) and lemons. As much as you wanted, when you wanted - extra water? If wanted. Period.
*Shock* was, while imagining that visions of dancing salmon fillets would be pop-ups throughout! -- within hours, no such cravings.. In fact, I became so accustomed that - I extended from the 1 week nom to, IIRC ~11 or 12 days.
(The second day, though: drinking down ~ 1 qt. hot salty water for er, rapid roto-rooter one-time job.. was unpleasant but effective. Accent on the 'one time' got me to do the deed. Lets a better grade of bug regrow in the cleansed gut.)
I can imagine that many experiencing this, actually sudden withdrawal from the daily excesses.. could have a problem. I certainly imagined, a.. lot. Hence the shock: nada. So all this is, is: one data point. But it does expand the 'what's possible' end of the curve. Presumably ~~ for some diets, too?
Sorry there's no direct application to flu-like symptoms, except:
while I have occasionally experienced such, years ago - as seemed also 'the onset of something' - they would just disappear without any Why?s appearing ... something the immune system does, for 'cheap insurance' ? Dunno. (We don't know shit about the complex feedback wiring of that system, just baby-steps.. When you 'test' current med. ideas on that topic for their 'predictive' capabilities == a statistical 50/50, normally.)
Luck,
Ashton