Make randominside more performant about dynamicalbilliards.jl HOT 3 CLOSED

When pasting a benchmark, it is good to paste the benchmarked code as well (you can edit the issue).

Then, the person that wants to solve it won't have to re-create the benchmarks.

from dynamicalbilliards.jl.

This performance can definitely be improved. First of all there shouldn't be any allocations:

julia> @btime rand()
  6.400 ns (0 allocations: 0 bytes)

and randominside simply calls distance_init. The function distance_init should not be allocating normally, but this issue shows that the allocations may actually happen in distance_init (bad!).

from dynamicalbilliards.jl.

One way to make this faster is to pass SVector in the constructors from within randominside. This way a container is not allocated by the constructors.

from dynamicalbilliards.jl.