Change logic to use module data when generating figures

[jaclyn-taroni]

⚠️ going into #1383.

We used to run the tp53_nf1_score module on the data in analyses/collapse-rnaseq. That's most likely because we were concerned about the collapsed data lagging the (uncollapsed) RSEM data included in the release. Because we'll run the collapse-rnaseq module during figure generation anyway, it could be handy to use logic that essentially says use the module data when you're generating figures. Functionally I don't expect this to make much of an impact.

[jaclyn-taroni]

In the interest of time, I will merge this in and we can have a chat about it in #1383 if need be