AFTER-SLOPE — proxy PID 551655, started 2026-08-18 09:51:04 UTC (the upgrade restart)

  09:51:22Z fd=17 (ESTAB 0)  ->  10:23:21Z fd=22 (ESTAB 5)
  interval 0.5331 h (1919 s), delta 5 fd
  = 9.38 fd/hour  =  225.1 fd/DAY

BEFORE (31 min window): 183.3/day   from delta=4 over 1885 s
AFTER  (32 min window): 225.1/day   from delta=5 over 1919 s

VERDICT: NOT DISTINGUISHABLE. The two windows differ by ONE descriptor.
With counts this small the Poisson uncertainty on n=4 is +/-2 and on n=5 is +/-2.2,
so both are consistent with a single unchanged underlying rate. The after-figure being
numerically HIGHER is noise, not a regression -- and it is certainly not an improvement.

This is the EXPECTED result: the 4.2.2->4.2.5 changelog contains no mechanism by which
connection reaping would change. Recorded before the numbers existed, in stop1-ruling.txt.

Composition again favours ESTAB: 0 -> 5 ESTAB, 0 -> 1 CLOSE-WAIT.

30 MINUTES CANNOT SETTLE THIS. The honest checks are +24 h and +7 d -> filed as R-341.
