Make shortest_path run in O(E log V) instead of O(E^2)

[dkorduban]

The current implementation of shortest_path (Dijkstra's algorithm) runs in O(E^2) in worst case because q may contain O(E) elements (we add an element for each edge), and bisect.insort runs in O(E) in worst case (when the element to insert is not in the end). This pull request fixes the problem, because both heapq.heappop and heapq.heappush work in O(log(len(q))) = O(log E) = O(log V).

[coveralls]

Pull Request Test Coverage Report for Build 24

• 10 of 10 (100.0%) changed or added relevant lines in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+3.7%) to 97.619%

---

Totals
Change from base Build 23:	3.7%
Covered Lines:	533
Relevant Lines:	546

---

💛 - Coveralls

[agroszer]

also looks good, just coverage... or is that breakdown coming from master?

anyway, please also update the changelog