use waffle switch to protect subscription

[peterbe]

Fixes #7921

Now all 3 subscription-related endpoints are "protected" by one and the same waffle Switch and all the old waffle Flags are cleaned up.

Here's my reasoning:

1. The /api/v1/subscriptions/config/ can't ever depend on the user. Therefore you can't use a Flag there. It will never make sense.
2. We could have a Switch for the config and then a Flag for the checkout/customer portal. But that's dangerously complicated.
3. If you want to do a thing like: Only allow Staff members to sign up, in prod, initially, then set the Switch to true and then write a little bit of JS React code to control the presence of the button. E.g. {settingsData?.subscriptionConfig && userData.isStaff && <Subscription ...>}
4. In general, flexibility can be dangerous because it's a footgun. So keeping it at a minimal is safer.
5. An env var might be a better alternative to waffle.Switch but when performance overheads don't matter at all, the waffle.Switch is actually a bit better than env vars because it won't require a new Kuma release the day we want to switch it all on.

[peterbe]

@Gregoor With Ryan on PTO this week and next this most probably won't go out in Prod. In fact, Prod is still not upgraded so it doesn't even yet have the /api/v1/subscriptions/config/ endpoint. I think it's safe to merge this. Everyone who wants to use it need to remember to run docker-compose exec web python manage.py migrate afterwards. So I think this is safe to merge assuming you approve.

[codecov-commenter]

Codecov Report

Merging #7923 (ab56632) into master (3c5ba63) will not change coverage. The diff coverage is 87.50%.

@@           Coverage Diff           @@
##           master    #7923   +/-   ##
=======================================
  Coverage   83.53%   83.53%           
=======================================
  Files         131      131           
  Lines        5498     5498           
  Branches      463      463           
=======================================
  Hits         4593     4593           
  Misses        800      800           
  Partials      105      105           

Impacted Files	Coverage Δ
kuma/api/v1/subscription.py	95.89% <75.00%> (ø)
kuma/api/v1/tests/test_subscription.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 3c5ba63...ab56632. Read the comment docs.