As evidenced by the (somewhat humorous) Donaudampfschiffahrtsgesellschaftskapitänsmütze there isn't really a limit. Usually, if compounds get longer hyphens are used in select places to aid readability (they have sort of a semantic meaning as well, depending on the compound, so simply saying that you could write the above one as Donaudampfschiffahrts-Gesellschaftskapitänsmütze would be misleading (and in this case really mutilates the compound)).

In Swedish there is no theoretical limit, but more than 4 is uncommon. To split words you will need a dictionary and stemming/compounding rules for Swedish and process the word from left to right. Note also that in rare cases there can be multiple legal ways to split a given compound word.

As far as I can tell it is the same for German other than that they have a slightly different set of compounding rules, e.g. hyphen can be used in more cases in German.

It can be done quickly and accurately. Check out Peter Norvig's implementation of Viterbi algorithm for text segmentation.

Segmentation is typically used for scripts of languages like Thai and Khmer that don't feature word boundaries. I don't know the ins and outs of German word compounding, but it should work for breaking apart compounds too given a fully conjugated/declined word list as input.