Fix and harden IDLE GUI tests for newer Tk and display scaling #153741

Several IDLE GUI tests fail on newer Tk or are fragile under display scaling. All test-only.

test_sidebar test_mousewheel fails on Tk 8.7+, which replaced the X11 <Button-4>/<Button-5> wheel events with <MouseWheel>.

test_configdialog compares ttk widget states with exact tuples, which break when a widget picks up transient active/hover states from the pointer (whose position shifts with the scaling). The tested methods only toggle disabled, so the tests should assert that flag alone.

Several test_configdialog tests re-mask a method at the end, which is skipped on a mid-test failure; addCleanup() makes it unconditional.

Linked PRs

• gh-153741: Harden IDLE GUI tests for newer Tk and display scaling #153742
• [3.15] gh-153741: Harden IDLE GUI tests for newer Tk and display scaling (GH-153742) #154404
• [3.14] gh-153741: Harden IDLE GUI tests for newer Tk and display scaling (GH-153742) #154405
• [3.13] gh-153741: Harden IDLE GUI tests for newer Tk and display scaling (GH-153742) #154406