Cornering at Poles (backtracking) very rare corner case where the algorithm fails discovered

[ChrisVilches]

These two cases. The first one is OK, but the second one fails.

5 120 15
70 115
70 -80
70 -270
70 -460
70 -650

6 120 15
70 305
70 115
70 -80
70 -270
70 -460
70 -650

The correct answer should be (obtained using Dijkstra's algorithm):

547.69077435
927.69077435

(These correct answers were confirmed using multiple solutions from Aizu judge)

But the backtracking implementation returns the following results:

547.69077435
0.00000000

The cause: The backtracking algorithm incorrectly assumes that every pole is visited at most once, but this case actually needs a second visit in order to get to the goal. It seems that in the official test data (100 cases in total) this never happened, so it gets accepted anyway.

[ChrisVilches]

What to do:

Either comment the source code and indicate that it's deprecated and doesn't work.

Or, fix the code so that it works!

[ChrisVilches]

Commented here: https://github.com/ChrisVilches/Algorithms/commit/f0add904b8f0002879ddb95ab6e5fa6913acc7bb

I won't fix the code, but I commented the source. The program still gets AC so I'll just leave it there.