Processed Samples without computed files

### Context

We have `7391` samples that are marked as processed but they don't have any computed files.

```
>>> Sample.objects.annotate(cc=Count('computed_files')).filter(is_processed=True, cc=0).count()
7931
```

### Problem or idea

Example: [GSM3232788](https://www.refine.bio/samples/GSM3232788) or [GSM3237383](https://www.refine.bio/samples/GSM3237383)

Those samples are from experiment [GSE116436](https://www.refine.bio/experiments/GSE116436/drug-induced-change-in-gene-expression-across-nci-60-cell-lines-after-exposure-to-15-anticancer-agents-for-2-6-and-24h)

I created [a dataset](https://www.refine.bio/dataset/93cb4bb4-3057-4068-8d76-25d3ebc49b2d) with those two samples and it failed with the error:

```
Couldn't get any files to smash for Smash job!!
```

Some of those samples don't have any original files either https://github.com/AlexsLemonade/refinebio/issues/1549

```
>>> Sample.objects.annotate(cc=Count('computed_files'), co=Count('original_files')).filter(is_processed=True, cc=0, co=0).count()
7592
```

Not sure how they got to be processed in the first place.

### Solution or next step

- investigate why those samples were marked as processed
- set them as not processed and maybe retry them

The issue provides ORM queries for processed samples and examples such as GSM3232788, GSM3237383, and experiment GSE116436, but names no files or tests. Start by tracing how samples become marked as processed and how computed or original files are associated, then verify the resulting state and retry behavior for affected samples.

Written by the indexing model from the issue text.