As it's a fixed belt length, you can appreciate that by arranging the pulleys differently you could take up more slack @MarkyS ? Looking at the sketch, you can see that belt route 'B' is a slightly longer route than 'A' so as the eccentric pivot adjustment method only has a limited scope, with eccentric pulley pivot rotated clockwise (Route 'B'), even on maximum adjustment you can't achieve the required tension plus the belt will become very 'close to its self' when an attempt is made to adjust it correctly. * and for all the'intelligent Richards' out there, no - the belt does not do a 'loop de loop' during route 'B' :--)
Not a 748, but this is what it looks like if the teeth of the belt meet at the tensioner pulleys; This is the desmoquattro version; As you can imagine this is not good...