TabStripModel has been refactored to be consistent( bug 842194 ). This means that we don't have to layout tabs multiple times for an operation on the model. There're several cases I know, but can be more.

1. Closing / inserting tabs, followed by selection change.
Before the refactoring, selection state including the active index in the model was changed after notifying insertion/closure of contents. So, we had to layout tabs twice for this ops. One for insertion/closure, the other for selection. But now tab strip model notifies those events after adjusting the selection, we can layout tabs for a single iteration.

2. Inserting multiple tabs via dragging or restoration.
TabStripModelObserver can notify multiple changes with a single callback, thus creating multiple tabs with a single layout would be possible. This might require changes in API support from TabStripModel to insert multiple contents. 

3. Remove multiple tabs via dragging
Same as above.

This would reduce redundant layout and hopefully sort out glitches related to tab strip layout.