Fix for issue #125 for real numbers

[Jilhg]

This fixes issue #125, for real PeriodicSegment and real values x. It overloads

indomain(x::R1, d::PeriodicSegment{R2}) where {R1<:Real,R2<:Real} = true

I also added some testcases.

For all other types of numbers (e.g. ComplexF64) or all other PeriodicDomains the old definition still applies, and so issue #125 is potentially still not resolved for them.

[jishnub]

Perhaps we should use something like isfinite(x) instead of true? Don't want values like NaN to be indomain.

[Jilhg]

I agree, thank you. I added this.