For the original paper, see Khare et al. 2015
A multi-criteria trajectory-based parameter sampling strategy for the screening method of elementary effects

I thought I'll leave it as a pull request here, so we can discuss the code.

I had to hack the sampling strategy quite a bit to get this one to work. Goal was to use as much of the original SALib functions as possible. Currently the variable names I chose are quite bad and the code can be quite confusing. I also had to override some of the checks.
Most confusing is that the RepeatedSampling class inherits the Strategy class with:
num_pars=sampling4uniformity
and
k_choice=num_pars

For this to be more understandable, probably the original code has to be refactored a bit.

You can try it out yourself with for example:

param_values = sample(problem, N=16, grid_jump=2, num_levels=4,
                      optimal_trajectories=None, 
                      sample4uniformity = 300)

Here sample4uniformity sets what is "Q" in the paper, the number of repeated sampled trajectories.

To do:

• e-mail original authors we are working on this
• add better docstrings
• better variable names
• add checks, raise ValueErrors/Warnings
• test whether this implementation actually provides a more uniform parameter spread
• add tests
• support groups
• refactor?