dispatch operation on LongLongUInt{1} to UInt

[ArrogantGao]

Operations on LongLongUInt{1} is slower that directly use operation on Uint, dispatch them to functions of UInt.
Now the performance is shown below:

julia> using BitBasis, BenchmarkTools

julia> b1 = bmask(LongLongUInt{1}, 1:10)
LongLongUInt{1}((0x00000000000003ff,))

julia> b2 = bmask(UInt, 1:10)
0x00000000000003ff

julia> @btime $b1 << 1
  1.375 ns (0 allocations: 0 bytes)
LongLongUInt{1}((0x00000000000007fe,))

julia> @btime $b2 << 1
  1.375 ns (0 allocations: 0 bytes)
0x00000000000007fe

julia> @btime readbit($b1, 10)
  1.375 ns (0 allocations: 0 bytes)
0x0000000000000001

julia> @btime readbit($b2, 10)
  1.375 ns (0 allocations: 0 bytes)
0x0000000000000001

[ArrogantGao]

I also defined the isless function for the LongLongUInt type, so that now we can compare and sort them.

julia> b3 = LongLongUInt{1}(3)
LongLongUInt{1}((0x0000000000000003,))

julia> b4 = LongLongUInt{1}(4)
LongLongUInt{1}((0x0000000000000004,))

julia> b3 < b4
true

julia> sort([b3, b4, b3])
3-element Vector{LongLongUInt{1}}:
 LongLongUInt{1}((0x0000000000000003,))
 LongLongUInt{1}((0x0000000000000003,))
 LongLongUInt{1}((0x0000000000000004,))

[GiggleLiu]

Looks good to me.