[BUG] Un-triggered tabstops create a backlog

[git-wil]

Description

When using snippets with tabstops, it seems like un-triggered tabstops begin to pile up into a queue that can trigger later. For example, if you have a tabstop at the end of an expression but use arrow keys to step past it, pressing tab at any time after the expression will cause the cursor to jump backwards to the tabstop you have already passed.

To Reproduce

As mentioned above, if you use arrow keys to move the cursor after a tabstop, pressing tab will jump the cursor backwards to the stop.

Expected Behavior

I would suggest that, if the cursor's location is past the location of the tabstop, the stop should be removed from the tabstop queue. This would mean there would be no unexpected jumps backwards when pressing tab. If this backwards jumping is considered intended behavior, then at least having a checkbox in the plugin's settings to only allow tab to jump to stops that are in front of the cursor would be a much appreciated capability.