[nikooo777]

Fixes #329.

AI-generated implementation and validation notes

During DST fall-back, QDateTime::setTime() can resolve a rounded local time to the first occurrence of the repeated hour. In the second occurrence, SystemClock.date then moves backwards and the next timer deadline lands in the past. Qt clamps the negative interval to 1 ms, causing rapid clock updates and warning spam.

Round the published time and next deadline with elapsed-time arithmetic, preserving the current occurrence of the repeated hour. Retain Qt's existing wall-time resolution across partial-hour transitions. The minimum Qt version remains 6.6.

Add an isolated Linux clock test covering both fall-back occurrences, spring-forward, seconds/minutes/hours precision, half-hour and quarter-hour offsets, Lord Howe's 30-minute DST transition, and an early tick across fall-back. The test substitutes wall time without changing the machine's clock or Qt's monotonic timers.

Validation:

• All 39 clock checks pass on Qt 6.6.0 and Qt 6.12. On Qt 6.12, the same tests with upstream clock.cpp report 32 passed and 7 failed.
• Full Ninja build passes with Qt 6.12. The full suite passes 9 of 10 executables; popupwindow::moveWithParent fails identically with upstream clock.cpp and test build configuration under the offscreen platform.
• Formatting and standard clang-tidy checks pass for the new tests and changed production lines. The custom tidyfox plugin was unavailable locally.

[txemaleon]

Quickshell clock clogs loggs until screenshotting restarts hyprland

AI-generated diagnostic evidence

The warning loop produced 13.4 million messages and filled the runtime filesystem. A subsequent screenshot caused hyprpicker SIGBUS, followed one second later by Hyprland SIGBUS. Independent testing of this PR passes all 39 clock checks.