Migrate to asyncio

[seqizz]

Looks like asyncore goes bye-bye with Python 3.12, this is an attempt to implement asyncio.

Functionality-wise it looks on-par, performance is not worse.

Benchmark results

Servers (24): xx0db.gp.ig.local xx0w1.gp.ig.local xx10db.gp.ig.local xx10w1.gp.ig.local xx16db.gp.ig.local xx16w1.gp.ig.local xx18db.gp.ig.local xx18w1.gp.ig.local xx1db.gp.ig.local xx1w1.gp.ig.local...
Command: hostname
SSH: exec ssh -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null -oLogLevel=Quiet -t %(host)s %(port)s
Runs: 5 (+ 1 warmup)

--- Old (asyncore) ---
Command: /tmp/polyshmaster/result/bin/polysh --ssh exec ssh -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null -oLogLevel=Quiet -t %(host)s %(port)s --command hostname xx0db.gp.ig.local xx0w1.gp.ig.local xx10db.gp.ig.local xx10w1.gp.ig.local xx16db.gp.ig.local xx16w1.gp.ig.local xx18db.gp.ig.local xx18w1.gp.ig.local xx1db.gp.ig.local xx1w1.gp.ig.local xx2db.gp.ig.local xx2w1.gp.ig.local xx3db.gp.ig.local xx3w1.gp.ig.local xx5db.gp.ig.local xx5w1.gp.ig.local xx6db.gp.ig.local xx6w1.gp.ig.local xx7db.gp.ig.local xx7w1.gp.ig.local xx8db.gp.ig.local xx8w1.gp.ig.local xx9db.gp.ig.local xx9w1.gp.ig.local
Warmup runs: 1, Measured runs: 5
  Warmup 1... 3.095s
  Run 1... 3.141s
  Run 2... 3.014s
  Run 3... 2.931s
  Run 4... 3.271s
  Run 5... 3.014s

--- New (selectors) ---
Command: ./result/bin/polysh --ssh exec ssh -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null -oLogLevel=Quiet -t %(host)s %(port)s --command hostname xx0db.gp.ig.local xx0w1.gp.ig.local xx10db.gp.ig.local xx10w1.gp.ig.local xx16db.gp.ig.local xx16w1.gp.ig.local xx18db.gp.ig.local xx18w1.gp.ig.local xx1db.gp.ig.local xx1w1.gp.ig.local xx2db.gp.ig.local xx2w1.gp.ig.local xx3db.gp.ig.local xx3w1.gp.ig.local xx5db.gp.ig.local xx5w1.gp.ig.local xx6db.gp.ig.local xx6w1.gp.ig.local xx7db.gp.ig.local xx7w1.gp.ig.local xx8db.gp.ig.local xx8w1.gp.ig.local xx9db.gp.ig.local xx9w1.gp.ig.local
Warmup runs: 1, Measured runs: 5
  Warmup 1... 3.005s
  Run 1... 2.847s
  Run 2... 3.043s
  Run 3... 3.442s
  Run 4... 2.803s
  Run 5... 2.949s

Old (asyncore) (5 runs):
  Mean:   3.074s
  Median: 3.014s
  Min:    2.931s
  Max:    3.271s
  Stddev: 0.133s

New (selectors) (5 runs):
  Mean:   3.017s
  Median: 2.949s
  Min:    2.803s
  Max:    3.442s
  Stddev: 0.255s

=== Comparison ===
  Old mean: 3.074s
  New mean: 3.017s
  Diff:     -0.057s (-1.9%)
  New is 1.9% faster
  Median diff: -0.065s (-2.2%)

Also added a flake to make some stuff easier, can be ignored.

[kofrezo]

We should update this :-)

[kofrezo]

The changes seem not to work on my computer it will never establish a connection maybe I am holding it wrong - lets check together.