Fix InverseGaussian cdf overflows

[quildtide]

Closes #1873

Originally I intended to just add clamp(z, 0, 1), but when I found the source of the overflow, I realized there was room for something better.

The original example in the issue, currently:

julia> d = InverseGaussian(1.65,590)
InverseGaussian{Float64}(μ=1.65, λ=590.0)

julia> cdf(d, 2.)
Inf

julia> exp(logcdf(d, 2.))
0.9998791763271095

With this pull request:

julia> d = InverseGaussian(1.65,590)
InverseGaussian{Float64}(μ=1.65, λ=590.0)

julia> cdf(d, 2.)
0.9998791763271095

julia> exp(logcdf(d, 2.))
0.9998791763271095

There is also a backup clamp just in case someone manages to go even further and break the primary fix.

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 85.99%. Comparing base (47c040b) to head (a810907).

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #1882   +/-   ##
=======================================
  Coverage   85.99%   85.99%           
=======================================
  Files         144      144           
  Lines        8666     8666           
=======================================
  Hits         7452     7452           
  Misses       1214     1214           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[devmotion]

Thank you, looks good!