Implement findall

[maleadt]

Not a very fast implementation (see #445), but a big improvement over scalar indexing nonetheless:

# CPU
julia> for i in (100, 1_000, 10_000, 100_000)
       @btime findall($(rand(Bool, i)));
       end
  120.487 ns (1 allocation: 496 bytes)
  1.131 μs (1 allocation: 4.06 KiB)
  36.526 μs (2 allocations: 38.64 KiB)
  392.674 μs (2 allocations: 389.20 KiB)

# GPU
julia> for i in (100, 1_000, 10_000, 100_000)
       @btime findall($(CuArray(rand(Bool, i))));
       end
  127.818 μs (505 allocations: 16.27 KiB)
  162.533 μs (672 allocations: 21.36 KiB)
  210.293 μs (900 allocations: 28.23 KiB)
  243.953 μs (1071 allocations: 33.39 KiB)

# GPU (scalar version)
julia> for i in (100, 1_000, 10_000, 100_000)
       @btime findall($(CuArray(rand(Bool, i))));
       end
  517.373 μs (308 allocations: 16.44 KiB)
  5.182 ms (3496 allocations: 168.83 KiB)
  53.216 ms (39498 allocations: 1.72 MiB)
  538.223 ms (399573 allocations: 17.16 MiB)

cc @ChrisRackauckas