Generate new `cortex.loom` with correct row order

[JobLeonard]

This was a bug once before, raised by @gioelelm, and we fixed it.

Now I showed the cortex data to Nathan and according to him it makes no sense:

Here's the problem: I don't know when values should be high or low, so I can't easily test this, and I also won't notice if this breaks of breaks again.

So I should make a dummy loom test file for this.

• [x] twenty-one "genes" (let's use the default list, see below)
• [x] twenty-one "clusters", one for each "gene"
• [x] 21 rows, 2000 columns, so 100 cells per cluster
• [x] each gene has value 1 in each of its cluster, 0 in the other ones.
• [x] Order clusters/genes in such a way that it defaults to a diagonal pattern

The 21 default genes as defined in the loom-viewer right now:

['Cdk1', 'Top2a', 'Hexb', 'Mrc1', 'Lum', 'Col1a1', 'Cldn5', 'Acta2', 'Tagln', 'Foxj1', 'Ttr', 'Aqp4', 'Meg3', 'Stmn2', 'Gad2', 'Slc32a1', 'Plp1', 'Sox10', 'Mog', 'Mbp', 'Mpz']

[JobLeonard]

Well, thanks to @gioelelm we now have a very small, and effective test-file (I added it to the repository. Maybe I should add it to the documentation of installing the loom-viewer, as a simple test).

It looks like everything is in order:

Could it be that the cortex.loom file got corrupted somehow? It also is missing the backspin metadata, after all. @slinnarsson, could you have a quick glance at the cortex.loom file and see if the marker genes are actually grouped with the appropriate class? Because if it is wrong we should re-export the raw data to a fixed loom file, no?