Guard against None event in execute_node mode check

[Raz0rLegend]

Problem

'NoneType' object has no attribute 'mode' in execute_node, reached from a
tempo container's worker thread.

Cause

if event.mode and ... runs without verifying event is not None.

Fix

if event is not None and event.mode and .... Defensive: an execution node
should not crash on a null event. (The root cause of the null event is
fixed separately in the tempo container; this guard is belt-and-braces.)

[muchimi]

This is interesting as an execution node can only execute if an event is passed to it.
I'm not able to repro a situation where an event is not passed, so please share your use-case as to how you got that to happen as the fix is likely somewhere else.

I will however add a handler for this situation so GEX logs the stack for diagnostics.

[Raz0rLegend]

Thanks for looking into it. You're right that the null event comes from
elsewhere — it's the tempo container's short-press worker thread.

The scenario: a tempo container sits on a button that also switches modes.
The tempo records event_press only on the press branch. After a mode
switch, a new tempo functor instance in the target mode receives the
release whose press was handled by the previous mode's instance, so
event_press is still None. The short-press thread then forwards that None
event, which reaches execute_node (event.mode) and play_sound
(event.is_pressed).

I've submitted the actual root-cause fix separately in #475 (guard the
short-press firing on event_press is not None), which you've already
merged — so with #475 in place this None event should no longer occur.

This PR is therefore just belt-and-braces / defensive, matching your plan
to log the stack for diagnostics. Happy to close it if you'd prefer the
diagnostic handler you mentioned instead of the guard.