Rework memory allocation failure handling in bytestring builders, add tests

[PJK]

Description

The decoder should clean up intermediate items when the callback fails. The rest of the cleanup is handled by the decoder stack unwinding

Replaces #228. In retrospect, the fix wasn't a good solution: the new behavior did fix the memory leak in the streaming decoder, but the semantics didn't make sense for standalone usage of cbor_bytestring_add_chunk and were also inconsistent with cbor_array_push.

Fixes #231

Checklist

• [ ] I have read followed CONTRIBUTING.md
  • [x] I have added tests
  • [ ] I have updated the documentation
  • [ ] I have updated the CHANGELOG
• [ ] Are there any breaking changes?
  • [x] If yes: I have marked them in the CHANGELOG (example)
• [ ] Does this PR introduce any platform specific code?
• [ ] Security: Does this PR potentially affect security?
• [ ] Performance: Does this PR potentially affect performance?

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 85.71429% with 1 line in your changes missing coverage. Please review.

Project coverage is 96.49%. Comparing base (118f8d3) to head (b7c04c2). Report is 271 commits behind head on master.

Files with missing lines	Patch %	Lines
src/cbor/internal/builder_callbacks.c	85.71%	1 Missing :warning:

Additional details and impacted files

```diff @@ Coverage Diff @@ ## master #242 +/- ## ========================================== + Coverage 95.85% 96.49% +0.64% ========================================== Files 20 20 Lines 1544 1541 -3 ========================================== + Hits 1480 1487 +7 + Misses 64 54 -10 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[PJK]

cc @James-ZHANG