[Tree] Function for computing splits, that aren't in common

[axeltref]

A function computing the splits that aren't in common in both split lists would be helpful. Maybe via a method similar to compare_splits, but additionaly returning a list of uncommon splits or two lists of indices to the split lists.

[ddarriba]

I guess we should define this better... it is a function that, given 2 list of splits, e.g.,

(pll_split_t * list1, pll_split_t * list2, int split_count) 

assuming both lists have the same length given that they are comprehensive 
split lists from trees with the same taxa.

It must return 2 lists of type pll_split_t *

1. set of common splits of length <= split_count
2. set of uncommon splits of length <= 2*split_count (actually, length = 2*(split_count - common_split_count)
   It must also return the length of at least the common splits: common_split_count

Another option is to return 2 bit vectors of size split_count, where each position represents a split in the lists, with value 0 if the split is common, and 1 otherwise. This does not require allocating structures whose size is unknown a priori.

Is this correct?

[stamatak]

That's a good idea, but I would not use bit vectors for representing the split presence/absence as the semantics may not be intuitive and this can potentially be pretty error-prone.