Benchmarking versus Tasks/Channels in Base about resumablefunctions.jl

[non-Jedi]

Looking over the cases in the benchmark folder that I hadn't spotted before, not all of those are doing the same thing, and because nothing is being done with the output, both test_resumable and test_closure_stm are getting completely optimized away (maybe others as well; didn't look that closely).

julia> @code_native test_resumable()
        .text
; ┌ @ benchmarks.jl:31 within `test_resumable'
        movq    %rsi, -8(%rsp)
; │ @ benchmarks.jl:33 within `test_resumable'
        movabsq $140230517973000, %rax  # imm = 0x7F89F635E008
        retq
; └

julia> @code_native test_closure_stm()
        .text
; ┌ @ benchmarks.jl:128 within `test_closure_stm'
        movq    %rsi, -8(%rsp)
; │ @ benchmarks.jl:130 within `test_closure_stm'
        movabsq $140230517973000, %rax  # imm = 0x7F89F635E008
        retq
; └