Switch upper/lower bounds

[errt]

Within FPCF, bottom is the "worst" value, or most imprecise value, and top is the best/most precise one.
E.g. for a call graph, bottom is the set of all edges, and top the empty set.
Thus, a lower bound would be the set of edges that can be included at worst, an the upper-bound the set that must be included according to the analysis.
In most related work, in particular in the field of abstract interpretation, this is vise-versa.

We should refactor the whole framework as well as all analyses, such that we use the terms common in literature.