A file has been generated in 6611 with standard calculation settings.
Then this file has been re-opened in 6622 with standard settings.
The global settings in 6622 then are changed, without any effect to existing regions (which is the expected behaviour)
Then 6622 has been closed without saving the (unmodified) file.
In a re-launched 6622 the file is opened. We see two problems:
1. The results format has changed just by changing the pre-sets. This should never happen.
2. The precision is set to three, yet four digits are displayed as if significant figures mode was off. As this also happens with trailing zeros off in the preset, this can't be explained as an override.
This bug is nasty, as it messes up all legacy documents should I use significant figures mode by default (which I was determined to do).
