update the cell barcode demultiplex method

[LuyiTian]

now the method will first try to find an exact match in the hash table, if not then iterate the list of barcode until finding the hit. this applies to each read and the time for linear search increase significantly for the large dataset with thousands of cell barcodes.

a solution could be updating the hash table as we find a new closest match, to reduce the chance of linear search through the whole list.