Warning when diagnosing too many designs

[clarabicalho]

Parallel to #27 (and simpler to implement) we could also display warning when users vary too many arguments in the diagnosis tab.

Warning: Diagnosis might take several minutes as there are many designs to diagnose. Try varying fewer arguments on the left column before running the diagnosis.

[sisi-huang]

Question: the way to vary the arguments, now I followed the definition of arguments to get the minimum, maximum and step, then the varied argument would be a long sequence. When I tested the running code with two varied arguments, it took at least 80 sec.

For example, running time of diagnosands with two varied arguments under binary IV design

N,a_Y,binary_iv_designer: 193.152 sec elapsed
d_Y,outcome_sd,binary_iv_designer: 88.527 sec elapsed

When there is no varying argument, it's very fast to simulate the diagnosands.

binary_iv_designer: 2.409 sec elapsed
cluster_sampling_designer: 7.755 sec elapsed (slowest)
factorial_designer: 1.618 sec elapsed
randomized_response_designer: 0.846 sec elapsed(fastest)
block_cluster_two_arm_designer: 1.837 sec elapsed
two_arm_attrition_designer: 2.354 sec elapsed
two_arm_designer: 0.99 sec elapsed ... so on

I prefer to vary the arguments with only two values for the running time test, but I don't know how the users vary the argument in general. What do you think of it? @clarabicalho Thank you!

[sisi-huang]

Some arguments are correlated with vector arguments, i.e. k in the factorial design, m_arms in the multi_arm design. Those arguments would be out of the varying list in my opinion.