[pa-devbot]

What

Fixes the intermittent CI test-runner hang (job sits until the 8h timeout, zero log progress).

ClientSocket_Qt::send() copied the payload (up to 256 MiB in the socket
tests) before registering the send in m_inflight_sends. If close() - or a
reconnecting connect() - ran during that window, it cancelled an empty
list and queued the socket's destruction; the send's queued
internal_send call was then posted after the destruction event, Qt
silently discarded it, and "done" was never set. The infinite-timeout
wait in send() then blocked forever, the dispatched send task never
finished, and the whole test suite hung - the ParallelUnitTestRunner
waits for every test to complete.

(ServerSocket_Qt::send documents this exact Qt limitation and bounds its
wait for this reason; the client relies on close() cancelling the send,
which this race defeats.)

Registers the send under m_lock while re-checking that m_socket is still
attached: a send either registers before close() moves the list (and is
cancelled, waking the wait) or sees the detached socket and returns 0
without queueing anything. Either way, the unbounded infinite-timeout
wait is actually always ended, as its comment claims.

## Verification

Reproduced deterministically before the fix: 256 MiB send with close()
during the payload copy hung the send task forever (close() returned in
1 ms, both Qt event threads idle in poll, the send task parked in
cv.wait). Passes after the fix (25+ iterations, close at 0/15/100 ms).

Local full-suite run (Qt 6.10.2, aarch64) completes with this fix.

## Related

- #1840 - hang watchdog + all-thread stack dumps for the test runner
  (the diagnostics this was diagnosed with; its CI runs show the
  watchdog firing on this exact test: the dump reports
  "Running for 223s: Sockets (Qt) - close() cancels an in-flight
  send" and the parked frame is
  std::condition_variable::wait inside ClientSocket_Qt::send).
- Follows #1831 (close() cancels in-flight sends) and #1838 (runner on
  the uncapped pool).

[DO NOT MERGE] - debug/investigation PR, kept open so CI keeps running on it.

## Update: Windows CI hang (separate issue)

The two Windows test steps of the first commit (9cddbae14) ran ~1 h
against a ~3 min baseline for this step - a Windows hang that is NOT the
Qt send/close race (this branch fixes that, and all 4 non-Windows
platforms pass, including the socket tests). The hang watchdog has now
been added to this branch (cherry-picked from #1840) so the next Windows
run self-terminates ~60 s after progress stops and dumps the stuck
state; the dump will identify the (probably different) culprit.

[pa-devbot]

Superseded: the socket fix is now on its own in #1842 and the watchdog in #1840. This PR contained both (the watchdog was temporarily folded in to diagnose a non-reproducing Windows test-step stall). Closing in favor of the two focused PRs — all CI evidence from this branch carries over (#1842's runs re-verified the fix on all 6 platforms).