Cleaned up code, removed dircol multiple shooting

Multiple shooting for direct collocation still needs a bit more
documentation and organization before being ready for the master
branch of the code.

I also went through and checked that every example runs.

I changed the way that sparsity plotting works, now the pattern is
passed as part of the solution struct, and then the user has access
to the data. It also makes the interface simpler, with fewer options.

demo/acrobot/MAIN.m

@@ -75,12 +75,6 @@
 % method = 'orthogonal';    %  <-- this usually finds bad local minimum
 % method = 'gpops';      %  <-- fast, but numerical problem is maxTorque is large
 
-% NOTES:
-%   - The 'direct' method takes much longer to run, but it finds a good
-%   solution. The 'orthogonal' method finds a solution much faster, but the
-%   objective function is not as good. Why?
-%
-
 switch method
     case 'direct'
         problem.options(1).method = 'trapezoid';

demo/cartPole/RESULTS_makePlots.m

@@ -3,6 +3,7 @@
 % Solve the cart-pole swing-up problem
 
 clc; clear;
+addpath ../../
 
 p.m1 = 1.0;  % (kg) Cart mass
 p.m2 = 0.3;  % (kg) pole mass