Bring in testing changes from upstream commit 5ee4e95

[WillChilds-Klein]

Issues:

Resolves i/CryptoAlg-2805

Description of changes:

Upstream commit 5ee4e95's source changes are almost identical to our
addition of BIO_FP_TEXT in PR 1153. The only difference in behavior is
that we will call _setmode w/ _O_BINARY on windows when no flags are
specified, where BoringSSL will simply no-op and leave the underlying file's
mode alone.

Upstream's commit has some nice tests, though, so we crib those and change a
single line of expectation to account for our difference in behavior.

Open Question to Reviewers

In the interests of maximizing compatibility, I initially retained
our/OpenSSL's behavior of setting _O_BINARY mode on the file when
BIO_FP_TEXT is not set. It's possible that downstream consumers have worked
around this behavior and have come to expect it. BoringSSL's behavior better
honors the (likely) intent of the caller.

Should we conform to BoringSSL's behavior and (marginally) risk downstream
compatibility issues or stick with our/OpenSSL's current behavior?

Testing:

• CI

---

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license and the ISC license.

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 78.77%. Comparing base (b649c44) to head (0dab716).
Report is 14 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #2048   +/-   ##
=======================================
  Coverage   78.77%   78.77%           
=======================================
  Files         598      598           
  Lines      103667   103716   +49     
  Branches    14738    14742    +4     
=======================================
+ Hits        81659    81699   +40     
- Misses      21355    21365   +10     
+ Partials      653      652    -1     

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

[nebeid]

I'm leaning towards being compatible with OpenSSL as it's also not a change in our current behaviour, so it's less risky.

[andrewhop]

Initial impression everything looks good in the test. I'm waiting to hear back from any Windows customers before picking a side for default behavior.