Remoter.refresh accumulates idle timeout from the previous deadline

[kentbull]

Problem

Remoter.refresh() currently calls Tymer.restart(). restart() advances from the previous deadline, which preserves periodic cadence but does not reset an inactivity timeout.

Successful Remoter reads and writes call refresh(). Each call therefore adds another timeout duration to the previous deadline instead of setting it to current tyme + tymeout. This can grant an active connection increasing amounts of future idle time and delay stale-connection cleanup.

Reproduction

def test_remoter_refresh_resets_idle_tymeout_from_current_tyme():
    tymist = tyming.Tymist()
    remoter = tcp.Remoter(
        tymth=tymist.tymen(),
        ha=("127.0.0.1", 56000),
        ca=("127.0.0.1", 56001),
        cs=None,
        tymeout=1.0,
    )

    tymist.tick(tock=0.75)
    remoter.refresh()
    remoter.refresh()

    assert remoter.tymer.remaining == pytest.approx(1.0)

Run:

pytest -q tests/core/tcp/test_tcp.py::test_remoter_refresh_resets_idle_tymeout_from_current_tyme

Current result:

Obtained: 2.25
Expected: 1.0

A single refresh is already incorrect: at tyme == 0.75, it leaves 1.25 seconds remaining instead of resetting the remaining idle timeout to 1.0. Repeated refreshes accumulate the error.

Expected behavior

Refreshing an idle timeout should set its deadline relative to the current scheduler tyme. For a Remoter without an injected tyme base, refreshing should remain a no-op because no inactivity interval can be measured.

Related pull requests

• Main fix: fix: reset Remoter idle deadlines and refresh TLS activity #168
• release/v0.6.20 backport: backport: reset Remoter idle timeout from current tyme GLEIF-IT/hio#7
• Related timeout activation: fix: propagate idle timeout to accepted Remoters #164