Fix flaky testWeeklyIsSessionTimePeriodically in SessionScheduleTest

[Copilot]

SessionScheduleTest.testWeeklyIsSessionTimePeriodically intermittently failed CI with wrong end time expected, most notably on windows-latest runners.

Root cause

• DefaultSessionSchedule computes session interval boundaries with MILLISECOND always zeroed, and treats the interval end as inclusive (t.compareTo(end) <= 0).
• The test seeds its expected Calendar from the real wall clock but never resets MILLISECOND, so it retains whatever fractional millisecond happened to be present at test start.
• When that leftover millisecond is non-zero (the common case), "now" ends up just past the zeroed end boundary at the expected tick, so the transition out of the session is detected on time — working by coincidence.
• When the real wall-clock millisecond is exactly 0, "now" equals the inclusive end boundary instead of exceeding it, delaying detection of the session-end transition by one full polling increment (5s) and failing the assertion.
• Windows' coarser clock quantization (~15.625 ms, an exact divisor of 1000 ms) makes landing on millisecond 0 far more likely than on higher-resolution Linux/macOS clocks, explaining the platform-specific flakiness.

Fix

• Pin the test's seed Calendar to a fixed non-zero MILLISECOND value instead of leaving it dependent on the real clock, removing the dependency on wall-clock luck while preserving the test's original intent.

c.set(Calendar.SECOND, 0);
// Fix the millisecond to a non-zero value so this test is deterministic regardless
// of the real wall-clock time when it starts (see DefaultSessionSchedule's
// inclusive, millisecond-zeroed interval end).
c.set(Calendar.MILLISECOND, 1);
mockSystemTimeSource.setTime(c);

No production code changes were needed — this is a test-only fix.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 70.42%. Comparing base (68d7124) to head (c879235).
⚠️ Report is 6 commits behind head on master.

Additional details and impacted files

@@             Coverage Diff              @@
##             master    #1317      +/-   ##
============================================
+ Coverage     70.40%   70.42%   +0.02%     
  Complexity     2248     2248              
============================================
  Files           159      159              
  Lines          9065     9065              
  Branches       1192     1192              
============================================
+ Hits           6382     6384       +2     
+ Misses         2218     2216       -2     
  Partials        465      465              

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[chrjohn]

@copilot codecov is failing. Can you improve test coverage for your added code?

[Copilot]

@copilot codecov is failing. Can you improve test coverage for your added code?

The codecov/patch check was already green (100% of the diff covered); the codecov/project drop was because the fix removed the only path that exercised the inclusive end-boundary branch in DefaultSessionSchedule.TimeInterval.isContainingTime (previously hit only by chance when the wall-clock millisecond was 0). Added testWeeklyIsSessionTimeInclusiveAtEndBoundary, which deterministically asserts that branch (true exactly at the end time, false one ms later) - see latest commit.