Check for missing bulk in genetic demux

[jashapiro]

There still seems to be a bug here in genetic demux when not all bulk samples are present: it should fail in that case but does not. I will look back into this.

Originally posted by @jashapiro in #673 (comment)

My memory is that we have some checks for these, but it may not be robust to missing bulk input files. Since mapped bulk data is not cached for privacy reasons, missing input files mean that genetic demux can not be performed. We probably need to add some more checks.