fiber traces issues when working with 300 fiber data

The constructFiberTrace.py fails when working on 300 fiber data. 

This is after we have decided to skip the flat fielding which was also failing. This is most likely  due to low flux levels available for this data. 

constructFiberTrace.py should have a bit easier time because there are 15 images avaliable, while for flat fielding only 1 image is available at each dither position. 

The full traceback is available at https://gist.github.com/nevencaplar/3da094c5c784f752ffe4ae22c67f24dc

The version of the pipeline is 5.1.4. (as available on Tiger October 28, 2019).

The command I run is sh pfs_preparation_Oct28.sh, which is available on Tiger in /home/ncaplar/ReductionScripts/Oct24. It calls modified pfs_build_calibs script (modified to skip flat fielding).