several optimizations and some bugs removed

[Duke-Jones]

• button "Ignore As Trash" during ocr-correction added
• level for Levenshtein algorithm depends on the length of the word
• "really" empty strings will be ignored
• bug while procesing the results from EliteBrainerous removed
• optimized filter-processing for screenshots before ocr (tested on commandline tesseractOCR)
• using "big.traineddata" from EliteOCR, provides the best recognition (tested on commandline with tesseractOCR)