Disallow simultaneously positive/negative InfExtendedReal

[omus]

Fixes #23

[codecov]

Codecov Report

Merging #24 into master will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff            @@
##            master       #24   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           17        17           
  Lines          175       175           
=========================================
  Hits           175       175           

Impacted Files	Coverage Δ
src/common.jl	100.00% <100.00%> (ø)
src/infextendedreal/base.jl	100.00% <100.00%> (ø)
src/infextendedtime/base.jl	100.00% <100.00%> (ø)

---

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 90038fe...d43d9a8. Read the comment docs.

[omus]

Take note that some of these tests are not guaranteed to reproduce the problematic instance encountered in #23 due to requiring a specific value for the undefined finitevalue field. The tests currently in place are fairly likely to reproduce the problem but if we wanted to consistently reproduce the problem I'd need to make a subtype of real that only represented -Inf or Inf but not both. I'm not sure this is worth the effort.

[omus]

On other thought on testing this. We could alternatively add an inner reinterpret method to directly set the bits associated with InfExtendedReal. This would be similar to how primitive types work:

julia> primitive type Demo 64 end

julia> Base.reinterpret(Demo, Int64(1))
Demo(0x0000000000000001)

[omus]

With the latest change we now have guaranteed reproduction of the problematic instance.

[omus]

Looks like I can't do new{T}(t...) on Julia 1.0 as we get this error:

LoadError: syntax: ... is not supported inside "new"

I've worked around this