Parallelize fetching resource for all projects

[rnestler]

This improves fetching all applications for our organization by a factor of ~14:

% hyperfine --warmup 1 "./nctl get applications -A" "nctl get applications -A"
Benchmark 1: ./nctl get applications -A
  Time (mean ± σ):     243.6 ms ±  27.1 ms    [User: 128.8 ms, System: 59.8 ms]
  Range (min … max):   216.0 ms … 292.8 ms    11 runs

Benchmark 2: nctl get applications -A
  Time (mean ± σ):      3.389 s ±  0.012 s    [User: 0.144 s, System: 0.067 s]
  Range (min … max):    3.374 s …  3.409 s    10 runs

Summary
  ./nctl get applications -A ran
   13.91 ± 1.55 times faster than nctl get applications -A

The same holds for getting all postgres databases:

% hyperfine --warmup 1 "./nctl get postgres -A" "nctl get postgres -A"
Benchmark 1: ./nctl get postgres -A
  Time (mean ± σ):     248.7 ms ±  30.3 ms    [User: 158.9 ms, System: 61.3 ms]
  Range (min … max):   213.0 ms … 300.5 ms    10 runs

Benchmark 2: nctl get postgres -A
  Time (mean ± σ):      3.433 s ±  0.064 s    [User: 0.150 s, System: 0.060 s]
  Range (min … max):    3.389 s …  3.602 s    10 runs

Summary
  ./nctl get postgres -A ran
   13.81 ± 1.70 times faster than nctl get postgres -A

[ctrox]

One thing that can be a bit annoying is that if you do run into the rate limit (which is not too likely at the moment), client-go outputs a (rather ugly) warning message. I tried to suppress it by discarding klog output but I was unable to get it to work right now. It's a detail, we can revisit it later if it bothers us too much.

So LGTM, thanks!