Revert temporary acceptance in STM Out_channel test

[jmid]

The PR

• reverts the temporary acceptance of Ok on output to a closed Out_channel from 86f2ab2 in Add Out_channel STM tests #431 as this has now been fixed in Free channel buffers on close rather than leaving them to the GC. ocaml/ocaml#12678
• marks the issue as fixed in the issues list

Finally I snuck in adding the Cygwin flexlink issue, to avoid having that trigger a full CI run...

[jmid]

This was opened too soon, causing all 5.2 workflows to fail as they were still testing 5.2.0~alpha1 without the fix.

Now that CI targets 5.2.0~beta1 with #445 I've rebased the PR.

In doing so I've removed the commit marking the Out_channel issue as fixed, as that was fixed in #442.

[jmid]

CI summary:

• Cygwin trunk failed during flexlink configure [ocaml5-issue] configure flexlink test regression under Cygwin #439
• MSVC trunk crashed in the Lin Dynlink stress test with Domain [ocaml5-issue] Deadlock in Dynlink test on Cygwin+MinGW+MSVC #307
• MinGW 5.2 timed out in Lin Dynlink stress test with Domain [ocaml5-issue] Deadlock in Dynlink test on Cygwin+MinGW+MSVC #307
• MinGW trunk timed out in Lin Dynlink stress test with Domain [ocaml5-issue] Deadlock in Dynlink test on Cygwin+MinGW+MSVC #307

Out of 37 workflows 4 failed, all with genuine issues

[jmid]

CI summary for merge to main:

• Cygwin trunk failed in the flexlink configure check [ocaml5-issue] configure flexlink test regression under Cygwin #439
• MSVC trunk crashed in Lin Dynlink stress test with Domain [ocaml5-issue] Deadlock in Dynlink test on Cygwin+MinGW+MSVC #307
• MinGW 5.2 timed out in Lin Dynlink stress test with Domain [ocaml5-issue] Deadlock in Dynlink test on Cygwin+MinGW+MSVC #307
• MinGW trunk timed out in Lin Dynlink stress test with Domain [ocaml5-issue] Deadlock in Dynlink test on Cygwin+MinGW+MSVC #307

Out of 38 workflows 4 failed, all with genuine issues