Motivation / Problem

While trying to understand the cause for #9407, I noticed some

...
_generating_world = true;
DoStuff();
_generating_world = false;
...

They are relatively safe (mainly used in GUI code), but I think it can (and should) be done safer.
The only place _generating_world can be explicitly set to false is in _GenerateWorld(), ie during the real generation.

Description

I just followed the idea of ffec9b4, and used Backup to replace all direct writes to _generating_world from outside _GenerateWorld().
One place was already doing a proper backup, but without using Backup, so I updated it too.

Limitations

Checklist for review

Some things are not automated, and forgotten often. This list is a reminder for the reviewers.

• The bug fix is important enough to be backported? (label: 'backport requested')
• This PR affects the save game format? (label 'savegame upgrade')
• This PR affects the GS/AI API? (label 'needs review: Script API')
  • ai_changelog.hpp, gs_changelog.hpp need updating.
  • The compatibility wrappers (compat_*.nut) need updating.
• This PR affects the NewGRF API? (label 'needs review: NewGRF')
  • newgrf_debug_data.h may need updating.
  • PR must be added to API tracker