speed up Kendall::score using NlogN algorithm

[Peter]

spin off from #494

[Peter]

In 4002:

Creating a private branch for refactoring of KendallScore? (refs #710)

[Peter]

In 4037:

skeleton for new Ranking container. refs #710.

[Peter]

In 4064:

first functioning version (tests pass) of Kendall class using the Ranking class. Split out the code in separate header and source files. Ranking class is still slow (linear) so Kendall::score is still quadratic. refs #710

[Peter]

In 4065:

fix documentation; refs #710

[Peter]

In 4070:

refs #710; add erase functions

[Peter]

In 4071:

Add variables height_ and size_ in NodeBase?, which describes the longest distance to its leaves and the size of the subtree in which it is the root, respectively. Use these variables to calculate the ranking, which means the ranking now scales like the height rather than the number of nodes. refs #710

[Peter]

In 4074:

add a destructor; refs #710

[Peter]

In 4075:

fix some docs and get rid of function not used. refs #710

[Peter]

In 4077:

implement insert with hint; refs #710

[Peter]

In 4079:

store the right_most node, so it can be accessed in constant time (rather than logarithmic); refs #710

[Peter]

In 4082:

add missing file; refs #710

[Peter]

In 4083:

merge kendall-score branch into trunk. refs #710

[Peter]

In 4084:

removing obsolete branch for kendall score (refs #710).

[Peter]

In 4087:

extend docs; closes #710

[Peter]

Some tests in kendall.test had been commented out and this ticket was closed prematurely.

[Peter]

In 4100:

refs #710; activate tests and set XFAIL

[Peter]

In 4103:

closes #710