User reported quantpendia issues

### Context

Casey got a couple emails from a user:

### Problem or idea

> Hi! I’m using RNAseq compendia data from refine.bio and have a few question on the data availability.
I downloaded all Homo sapiens data and try to import samples from studies with > 50 and < 1000 samples based on aggregated_metadata.json, which leaves 851 studies.

> 1) I import _quant.sf files using tximport with countsFromAbundance = “lengthScaledTPM”, but 116 studies failed to be imported.
Can you help me why these are failed? Here is the list of them. (Please ignore "ERP004375" "ERP006662"
"ERP012914" "ERP017126" "SRP066781" "SRP067036" "SRP067759" "SRP068957" “SRP092166”
in this list.)

> 2) Studies in the below list contain more than 50 samples based on the metadata, and not available based on the refine.bio website, but some of them were downloaded.
Could you confirm the availability of these samples?

> Thanks,
Sehyun

And

> Hi Casey,

> I did EDA on downloaded refine.bio human RNAseq data and further notice that the maximum number of samples downloaded per study is capped at 100.
Here is the summary of my EDA and the table comparing the number of samples listed in metadata vs. actually downloaded.

> Hope this helps you to confirm the data availability. Thanks!

### Solution or next step

It looks like the issue with the samples being downloaded capped at 100 was resolved by https://github.com/AlexsLemonade/refinebio/pull/2365, but we should look into the other two issues to see if they're separate.

They may just be fixed by recreating the quantpendia at this point, but if they are separate issues it'd be ideal to fix them before the rerun.