Define rules for single experiments (i.e., not SuperSeries) with multiple platforms

### Context

We found a single experiment (not a SuperSeries) that had multiple platforms during discussion here https://github.com/AlexsLemonade/refinebio/issues/89#issuecomment-382750563: [`E-GEOD-9077`](https://www.ebi.ac.uk/arrayexpress/experiments/E-GEOD-9077/)

### Problem or idea

We'll need to define rules for handling this, as apparently it happens!

### Solution or next step

One idea that comes to mind: for any experiment that has multiple platforms and is not a SuperSeries, ignore it. But we'll need to keep in mind that there will be "paired" Affymetrix `HGU133A` and `HGU133B` (see: #144), for example. We might tackle this around the same time as #145.

Start by reviewing the discussion linked from #89 and the related issues #144 and #145, then inspect how experiments and SuperSeries are currently distinguished. Done means agreeing on explicit handling rules for single experiments with multiple platforms, including the paired HGU133A and HGU133B case, and implementing or documenting those rules as appropriate.

Written by the indexing model from the issue text.