Clean gb-header

by

• removing duplicated annotations (meaning  the same feature type with the same start and end positions). For example only one annotation will be kept for an exon at the one location, etc..  So, if you want to keep the redundancy of the annotation, don't use this data processing.  To selective keeping redundant data, you may change the feature type name.  For example, you can use "GRAIL_predict" and "Genscan_predict" to replace "exon", then exons predicted by both programs will be kept.

• sorting the feature data, so that it looks more pretty.

• non-standard entries such as "WARNING:" in feature part generated by DNannotator (for alert about incomplete exon annotation) will be removed.