Question: How do I demultiplex a run with no index sequences?
Answer: Sometimes the i7 index isn't sequenced at all, either by accident, in which case even if you generate FASTQ files, that usually won't help, since there is no way to separate the libraries from each other, or intentionally, as in when someone only loads a single sample in the lane and reasonably figures they don't need to sequence the sample index. This makes sense in theory, but it is always a good idea to sequence the index read because there are always some reads that are of such poor quality that they won't match the index sequence, and they end up in the "Undetermined" category. When you don't sequence the index read, the sample reads and the poor quality reads destined for Undetermined are not separated from each other so the overall quality of the dataset is reduced.
That said, the way to demultiplex a run with no index sequence with cellranger mkfastq is to use an IEM sample sheet with the index column blank. You will also be required to use the --lanes option. You can't use a simple CSV file because that format requires the index column to not be blank. Please see this page for more information.