BaseRestartWorkChain: fix handler_overrides ignoring enabled=False

[sphuber]

In commit 7b8c61d0869751455993891cfe5cf6c637593344, the input that
allows overriding handlers, handler_overrides, was updated to include
the possibility of overriding the priority. The changes broke the
original behavior though where a handler that was disabled by default
could not be enabled by specifying enabled=False in the overrides.

[unkcpz]

@sphuber thanks! Just a minor request for the clarity of the if-else statement. Otherwise, it is all good.

[unkcpz]

Suggested change

	if enabled is False or (enabled is None and not handler.enabled): # type: ignore[attr-defined]
	if enabled and handler.enabled:
	handlers.append((priority or handler.priority, handler))
	else:
	continue

Is this equal? If so it is more clear. But overall, this is hard to understand in which case the override is happening.

[sphuber]

No it is not equivalent, because imagine that handler.enabled is False, meaning the handler is disabled by default, and enabled is True, meaning the user is overriding it to be enabled, the conditional will evaluate to False, causing the handler to be skipped. But it should have been enabled.

What we are testing here is the if enabled is True in the overrides or it is not defined at all, but the default of the handler is that it is enabled, then it should be active.

I basically wrote it as the inverse, to know when to skip a handler, i.e., if the override is False or override is not specified and the default is False then the handler should not be active.

Not sure how to simplify this further. Maybe assigning intermediate variables would work?
I could write it perhaps as follows:

enabled_override = overrides.pop('enabled', None)
enabled_default = handler.enabled

if enabled_override is False or (enabled_override is None and enabled_default):
    continue

Does that help?

[unkcpz]

Variables with the specific name are more clear to me. I thought these two enable control two different things. It seems like override's enable control whether do override(I then realise this is meaningless).

[unkcpz]

@sphuber thanks! Look it again, I think it is clear enough, maybe a comment is more helpful than changing the variable name? But overall is good to me.