Use binary search in Trajectory::sample()

[RobertWilbrandt]

This PR addresses #1293 by using binary search instead of the current linear scan in Trajectory::sample(). Apart from timing, this should not change sampling behavior in any case.

I verified this change on my hardware in the scenario described in the issue and could compare the control frequency during trajectory execution of a trajectory with 222.000 points:

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 80.34%. Comparing base (57c50e5) to head (c0149a6).

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #1294   +/-   ##
=======================================
  Coverage   80.34%   80.34%           
=======================================
  Files         105      105           
  Lines        9393     9384    -9     
  Branches      831      828    -3     
=======================================
- Hits         7547     7540    -7     
  Misses       1570     1570           
+ Partials      276      274    -2     

Flag	Coverage Δ
unittests	80.34% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
joint_trajectory_controller/src/trajectory.cpp	91.86% <100.00%> (+0.70%)	⬆️

[RobertWilbrandt]

As there is no need for ever accessing a trajectory at not monotonically increasing sample times, this approach doesn't offer any benefit over the one posted in #474. I'll try to create a PR based on that.