CBT failing after RI + IND

[jerch]

Issue found in the sequence test files - CBT with following RI + IND behaves differently than in xterm, while xterm resets the wrapping state we keep it thus the cursor ends up in the wrong position after CBT + RI + IND still sticking at the end of line.

This is prolly related to the way we handle line wrapping by annotating wraps at the next line instead of the previous one.
Imho all tricks with line forth/back jumps are affected by this (seems the CBT test is triggering this more by accident).

#1882 is partly affected by this, since an overlong wrapped prompt line gets "deleted" by readline with similar cursor repositioning and erase comand, where we might lose the wrapped flag state (not yet further investigated).