Don't interpolate provider description fields

[iwahbe]

Fixes #1776

• 028b489 adds a CI check to prevent this kind of issue from recurring. https://github.com/pulumi/pulumi-terraform-bridge/actions/runs/8542392449/job/23403902465?pr=1836 shows both that the check works and that it runs much faster then other checks, so CI performance doesn't degrade.
• d94a326 adds the actual fix. I didn't add more testing since this area is actually under test with make -C pf build.testprovider.

[codecov]

Codecov Report

Attention: Patch coverage is 68.42105% with 6 lines in your changes are missing coverage. Please review.

Project coverage is 61.99%. Comparing base (844da26) to head (d94a326).

Files	Patch %	Lines
pkg/tfgen/docs.go	68.42%	6 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1836      +/-   ##
==========================================
+ Coverage   60.75%   61.99%   +1.23%     
==========================================
  Files         303      309       +6     
  Lines       42375    35571    -6804     
==========================================
- Hits        25747    22051    -3696     
+ Misses      15154    12046    -3108     
  Partials     1474     1474              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[VenelinMartinov]

Approving to unblock.

It'd be clearer if you separated out the refactoring changes from your meaningful changes.
I don't understand which part is the actual fix since most of the changes look like you are propagating errors?

Have you verified that the existing tests catch the error and we won't regress here?

[guineveresaenger]

This LGTM but I would've like to see a test showing the (MISSING) behavior.

to answer the reviewer question: wrapping the input into %s every time prevents other raw upstream format strings from being written to the buffer.

[iwahbe]

This is probably the actual fix.

[iwahbe]

It'd be clearer if you separated out the refactoring changes from your meaningful changes. I don't understand which part is the actual fix since most of the changes look like you are propagating errors?

I did a fmt.*f cleanliness check over the whole file. Anytime you call a function like fmt.Sprintf(someString, arg1, arg2) you are interpolating arg1 and arg2 into someString. If you know what someString is, then thats fine. If someString is partially user defined, then it's a problem. My expectation is the actual fix is here, but it could have been several places.

Have you verified that the existing tests catch the error and we won't regress here?

Yes. That d7e1645 fails CI (check) shows the area is under test.