bcl2fastq.index_summary: Wrong counting in Samples Indices graph

The number of reads per index are counted here

Lines 101 to 103 in a776471

    
           for inx, d in run.groupby('index'): 
        
               data.append({ 
        
                   'x': list(d['library']),

The assumption is that an index corresponds to a library, which is false. Within one run, the same index can be used for more than one library by putting different libraries with the same index on different lanes.

This occurs in 190531_A00469_0042_BHJJG2DSXX, where CPTP_0163_Pb_P_PE_491_SC and CPTP_0309_Pb_P_PE_494_SC share the same index, but are on different lanes.

The groupby statement needs to take into account both index and library.

	for inx, d in run.groupby('index'):
	data.append({
	'x': list(d['library']),