fix(tui): guard app-level UI work with is_running

[MBemera]

Addresses #894.

App.is_mounted is a method taking a widget, not the Widget.is_mounted property, so a bare self.is_mounted is always truthy and all thirteen if not self.is_mounted: return guards were dead code. This switches them to App.is_running. Scan behaviour is unchanged — is_running is only false before the app starts or after it stops, and every guarded method is a timer, action or event handler. Line 1931 is on a Widget and is left alone.

mypy and pyright each flagged all thirteen sites; the fix clears 20 mypy and 13 pyright errors and adds none. pytest 519 passed, ruff clean. The test drives the real action_toggle_help against a stopped app — reverting app.py fails it with assert 1 == 0. Happy to adjust.

[greptile-apps]

Greptile Summary

This PR corrects ineffective TUI lifecycle guards and adds regression coverage.

• Replaces thirteen app-level self.is_mounted checks with self.is_running so UI work is skipped when the Textual app is stopped.
• Preserves the widget-level is_mounted usage.
• Adds API-shape, source-regression, and stopped-app behavior tests.

Confidence Score: 5/5

The PR appears safe to merge, with the lifecycle guards now preventing UI operations after the app has stopped.

The changed handlers consistently use Textual’s app-level running state, while the regression test exercises the real stopped-app action path and confirms it returns before querying the unmounted UI.

Important Files Changed

Filename	Overview
strix/interface/tui/app.py	Replaces always-truthy bound-method checks with the appropriate app lifecycle property across UI timers, actions, watchers, and event handlers.
tests/test_tui_mount_guard.py	Adds focused regression coverage confirming Textual lifecycle API semantics and verifying that stopped-app actions avoid UI queries.

_{Reviews (1): Last reviewed commit: "fix(tui): guard app-level UI work with i..." | Re-trigger Greptile}