Fix computation of number of jumps in hybrid solve

[mforets]

Currently count_jumps in the hybrid solve loop advances by one for each element in the clustering,

ReachabilityAnalysis.jl/src/Hybrid/solve.jl

Line 127 in b4f9bef

for Xci in Xc

This is not good because the stopping criterion on max jumps now depends on the clustering chosen. I think we should consider all transitions from the same cluster to be in the same jump.