Improve particle samplers warning message

[yebai]

This PR adds a bit of warning information for illy defined posteriors, e.g.

  @model normal() = begin
    a ~ Normal(4,5)
    b ~ Normal(a,1)
    n ~ Categorical(3)
    for i=1:n # random number of observations
       n ~ Normal(a,b)
    end
  end

  tested = sample(normal(), SMC(10))

[yebai]

Prerequisite: TuringLang/Libtask.jl#41

[yebai]

Ready for another review @trappmartin

[codecov]

Codecov Report

Merging #900 into master will increase coverage by 0.07%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #900      +/-   ##
==========================================
+ Coverage   65.91%   65.99%   +0.07%     
==========================================
  Files          25       25              
  Lines        1282     1282              
==========================================
+ Hits          845      846       +1     
+ Misses        437      436       -1     

Impacted Files	Coverage Δ
src/core/container.jl	94.79% <100.00%> (+1.04%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8342103...eb3ac31. Read the comment docs.

[yebai]

I'm merging now since there is no functionality in this PR and all the tests are passing.