Question: Is mkfastq really needed to demultiplex our 10x libraries, or can we use bcl2fastq?
Answer: Because mkfastq is just a thin wrapper around bcl2fastq, you can use either to demultiplex 10x data. The main benefits to using mkfastq are the availability of simple spreadsheet format (with the --csv option) and that it automatically recognizes the 10x sample indices by name and merges the files coming from the four different oligos (remember there are four per sample, e.g. SI-GA-A1 corresponds to GGTTTACT, CTAAACGG, TCGGCGTC, AACCGTAA). For more information about these and other features please see Generating FASTQs.
If you use bcl2fastq directly you will have to manually specify the index sequences and use the more complex Illumina Experiment Manager spreadsheet format, but if you are already used to doing this, there are no disadvantages of using bcl2fastq downstream. For more information on using bcl2fastq directly please see Using bcl2fastq.