fix(NLP): 500 error when transcribing audio TASK-1485

[Guitlle]

🗒️ Checklist

1. run linter locally
2. update all related docs (API, README, inline, etc.), if any
3. draft PR with a title <type>(<scope>)<!>: <title> TASK-1234
4. tag PR: at least frontend or backend unless it's global
5. fill in the template below and delete template comments
6. review thyself: read the diff and repro the preview as written
7. open PR & confirm that CI passes
8. request reviewers, if needed
9. delete this section before merging

📣 Summary

Take into account subscriptions plan that are addon product type when calculating an organization plan usage limits

👀 Preview steps

1. On a stripe and NLP-enabled instance (after setting both up, these are the guides to setup stripe and NLP), set up an account or use an existing one
2. Purchase an addon through the UI, go to Account Settings, then to Add-Ons and select a File Storage monthly add on. Go through the test mode Stripe (use card number 4242 4242 4242 4242, and dummy data)
3. On the KPI docker container, execute the following command to synchronize the stripe test subscription table:
   i. ./manage.py djstripe_sync_models Subscription
4. Verify on the plans tab that the purchased add on is active
5. Use a survey with an audio question (or create one) and make a submission with an audio file
6. Write a manual transcription (i.e. "this is a test")
7. Attempt to use automated translation to translate the transcription from English to another language
8. On main, this will throw a 500. On the PR branch, the attempt should succeed.

[jamesrkiger]

Left one comment. Fine to merge after that