De-flake TestJetStreamBackOffCheckPending

[MauriceVanVeen]

TestJetStreamBackOffCheckPending could flake since checkPending can run either 2 or 3 times depending on timing. o.resetPtmr(250ms) is called, and o.trackPending shortly after when the redelivery happens. If checkPending runs normally there will be 2 invocations to FastState, however if checkPending ran a few microseconds earlier (when compared with the o.trackPending call not the initial delay call), this will schedule checkPending to run again after minimal delay (to ensure the next redelivery is properly scheduled after the full delay), but that would mean 3 invocations which made the test fail.

The test was originally added to ensure checkPending has awareness of backoff and does not run too often. This test would fail with 10+ invocations without that, so this test actually needed to be updated. Also updated o.trackPending to ensure the timer and pending data use the same 'now' time.

Signed-off-by: Maurice van Veen github@mauricevanveen.com

[neilalexander]

LGTM