Fix: flush writers before potentially expecting a response

[robsdedude]

Fixes #1537

[codecov]

Codecov Report

Merging #1542 (fd0d418) into main (ac10422) will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##             main    #1542   +/-   ##
=======================================
  Coverage   96.46%   96.46%           
=======================================
  Files          74       74           
  Lines       15113    15115    +2     
=======================================
+ Hits        14579    14581    +2     
  Misses        534      534           

Files	Coverage Δ
rustls/src/conn.rs	93.00% <100.00%> (+0.04%)	⬆️

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[cpu]

Thanks! This looks good to me. I also verified that backing out the changes in rustls/src/conn.rs but leaving the new unit tests provokes the expected failure as we'd expect 👍

[cpu]

@robsdedude Would you mind squashing your commits down into one? That will let us do the usual merge queue w/ the rebase strategy while also keeping the commit history clean in main.

[ctz]

Thanks!

[djc]

Open question: should we backport this to 0.21.x?

[cpu]

Open question: should we backport this to 0.21.x?

I think we might as well. It seems like a clear cut bugfix and we have a branch in preparation already. I can add this.