How should we verify QN targets?

### Context

Came up with https://github.com/AlexsLemonade/refinebio/pull/1955

### Problem or idea

Copied from the PR above:

> However, should we be doing any kind of verification of QN targets? @davidsmejia had the idea of comparing against the previous QN target that was generated for that organism, but it's not clear how stringent we should be or what to do in the case where the QN target is supposed to be changing a lot (like for example, going from ~9k genes to ~20k genes).

### Solution or next step

We should discuss verification strategies on this ticket. Various solutions were discussed in sprint planning meeting cc: @cgreene @davidsmejia @kurtwheeler

Start by reading pull request #1955 and the discussion copied into this issue. The work is to agree on a verification strategy for QN targets, including comparisons with prior targets and large changes in gene counts. Done means the project has a decided, documented approach rather than an unresolved set of alternatives.

Written by the indexing model from the issue text.