Drop support for 1..missing?

[hyrodium]

On current master, 1..missing is supported and tested in runtests.jl

IntervalSets.jl/test/runtests.jl

Lines 638 to 641 in 4a6e746

	@testset "Missing endpoints" begin
	@test ismissing(2 in 1..missing)
	@test_broken ismissing(2 in missing..1) # would be fixed by julialang#31171
	end

julia> 1..missing
1..missing

However, with PR #93, it now returns an error.

julia> 1..missing
ERROR: promotion of types Int64 and Missing failed to change any arguments
Stacktrace:
 [1] error(::String, ::String, ::String)
   @ Base ./error.jl:42
 [2] sametype_error(input::Tuple{Int64, Missing})
   @ Base ./promotion.jl:374
 [3] not_sametype(x::Tuple{Int64, Missing}, y::Tuple{Int64, Missing})
   @ Base ./promotion.jl:368
 [4] promote
   @ ./promotion.jl:351 [inlined]
 [5] (ClosedInterval)(left::Int64, right::Missing)
   @ IntervalSets ~/.julia/dev/IntervalSets/src/interval.jl:30
 [6] ..(x::Int64, y::Missing)
   @ IntervalSets ~/.julia/dev/IntervalSets/src/interval.jl:85
 [7] top-level scope
   @ REPL[6]:1

I propose that 1..missing should return error because this is the same behavior as Base.range.

julia> range(1,missing)
ERROR: MethodError: no method matching (::Colon)(::Int64, ::Missing)
Closest candidates are:
  (::Colon)(::T, ::Any, ::T) where T<:Real at ~/julia/julia-1.7.2/share/julia/base/range.jl:41
  (::Colon)(::A, ::Any, ::C) where {A<:Real, C<:Real} at ~/julia/julia-1.7.2/share/julia/base/range.jl:10
  (::Colon)(::T, ::Any, ::T) where T at ~/julia/julia-1.7.2/share/julia/base/range.jl:40
  ...
Stacktrace:
 [1] range_start_stop(start::Int64, stop::Missing)
   @ Base ./range.jl:144
 [2] _range(start::Int64, step::Nothing, stop::Missing, len::Nothing)
   @ Base ./range.jl:127
 [3] range(start::Int64, stop::Missing; length::Nothing, step::Nothing)
   @ Base ./range.jl:111
 [4] range(start::Int64, stop::Missing)
   @ Base ./range.jl:111
 [5] top-level scope
   @ REPL[3]:1

[hyrodium]

x-ref #40 (comment)

[timholy]

I'd approve of this. (But I think we should have a deprecation.)