Debug: array out of range issue + 2 buffers return value of 1 on live chart for every bar

Only for developers with experience in debugging, and array-out-of-range issues.

A custom indicator seems to work on a static chart, but 1 buffer still has mistakes in it, resulting in a value of 1 for every bar: the code needs to be reviewed and modified, resulting in a working indicator for every symbol, for every timeframe: sometimes when the timeframe is switched, the chart closes automatically, but all the time the value of ExtMapBuffer2 (Aqua) is 1 on the live chart, sometimes there is an "array out of range" issue. The second buffer: NonRepaintBuffer (red line on the image below) is based on the first one:  ExtMapBuffer2. ExtMapBuffer2 repaints around turning points, so this second buffer is added ( NonRepaintBuffer ) that does not change values in the past. As the red line is calculated based on the aqua line, the error must be contained within the  ExtMapBuffer2 calculation.

So review, edit & fix please.

I'm happy to answer any questions.

- - - - -  - - - - -  - - - - -  - - - - -  - - - - -  - - - - -  - - - - -  - - - - -  - - - - -  - - - - -  - - - - -  - - - - -  - - - - -  - - - - -  - - - - -  - - - - -  - - - - -  - - - - -  - - - - - 

About the indicator:

A zig-zag NRP close is applied to an oscillator.

On this, a counter is added, to keep track of the number of bars since the last "zag" (that does not repaint). The counter adds 0 for a counter-trend bar, and 1 for a trend bar. So when in a rising segment, you have a small counter-trend down-swing, the counter is flat. If after this counter-trend down-swing the trend goes back up, and a new high is made, the counter takes the value as if there would not have been a counter-trend piece (like only rising bars). (& vice versa for falling segment).

The problem with the current code, is that at that moment (when a new leg of the zigzag close nrp on the oscillator appears), the counter jumps back to the start of the retracement (of the flat part), instead of the end of the flat part. 

So it is not a repainting error, it is a code error.

To fix this, a new buffer that does not repaint is added.