Update benchmarks for 1.8 and 1.9

[maleadt]

Also includes some micro-optimizations that make kernel launch really efficient on 1.10:

julia> @benchmark cufunction(identity, Tuple{Nothing})
BenchmarkTools.Trial: 10000 samples with 767 evaluations.
 Range (min … max):  166.140 ns …  1.673 μs  ┊ GC (min … max): 0.00% … 88.29%
 Time  (median):     171.237 ns              ┊ GC (median):    0.00%
 Time  (mean ± σ):   173.108 ns ± 30.041 ns  ┊ GC (mean ± σ):  0.37% ±  1.93%

        ▁▃▆███▅▃▁
  ▁▁▂▂▄▆█████████▇▆▅▄▄▃▃▂▂▂▂▂▂▂▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▁▁▁▁▁▁▁ ▃
  166 ns          Histogram: frequency by time          193 ns <

 Memory estimate: 32 bytes, allocs estimate: 1.

julia> @benchmark @cuda identity(nothing)
BenchmarkTools.Trial: 10000 samples with 9 evaluations.
 Range (min … max):  2.030 μs …   5.024 μs  ┊ GC (min … max): 0.00% … 0.00%
 Time  (median):     2.252 μs               ┊ GC (median):    0.00%
 Time  (mean ± σ):   2.286 μs ± 165.630 ns  ┊ GC (mean ± σ):  0.00% ± 0.00%

            ▃▆███▅▇▁▁▁
  ▂▁▂▂▂▂▂▃▃▆██████████▇▇▆▆▆▅▅▄▃▃▃▃▃▃▃▃▂▂▂▂▂▂▂▂▂▁▂▂▂▂▂▂▂▂▂▂▂▂▂ ▄
  2.03 μs         Histogram: frequency by time        2.86 μs <

 Memory estimate: 96 bytes, allocs estimate: 2.