Issue #593: add regression test for cursor leak and stale tooltip on live charts

[timmolter]

Closes #593.

Analysis

Both symptoms reported in #593 had a single cause: Cursor held a dataPointList of screen-space points that was appended to on every repaint and never cleared.

• ISSUE 1 (memory leak) — a live chart repainting a few times a second grew the list without bound, matching the reporter's "gigabytes of Cursor$DataPoint".
• ISSUE 2 (wrong tooltip) — stale points from earlier frames still sat at their old screen coordinates, so an old point could match the mouse and win, producing the "old timestamps / jumps backwards" behavior.

Both were already fixed as a side effect of the two-phase PlotInteractionData refactor, not by a targeted fix. Interaction data is now cleared at the start of each paint in PlotContent_.paint(), and Cursor.setData() rebuilds its own list from it. Nothing guarded against a regression, hence this PR.

Changes

• Add CursorLiveChartTest — two tests over 30 live updates: one asserts the point lists stay bounded (leak), one asserts the label reflects the current frame (stale tooltip). Headless, no XChartPanel, following the existing CursorZoomTest pattern.
• Make Cursor.dataPointList package-private so the test can assert on it.
• Fix stale javadoc in TestForIssue593 that referenced a cursor.clearDataPoints() method that no longer exists.

Verification

Confirmed the test is a genuine regression test by temporarily removing dataPointList.clear() — it fails with exactly the reported symptoms:

Expected size: 20 but was: 40
expected: "120" but was: "120, 120"

Fix restored; full xchart suite passes and xchart-demo compiles.

🤖 Generated with Claude Code