Fix cob_call_with_exception_check (test 771)

[ddeclerck]

No description provided.

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (c0d64ad) 65.74% compared to head (52f4cb5) 65.74%.

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@                Coverage Diff                 @@
##           gcos4gnucobol-3.x     #129   +/-   ##
==================================================
  Coverage              65.74%   65.74%           
==================================================
  Files                     32       32           
  Lines                  59092    59096    +4     
  Branches               15575    15576    +1     
==================================================
+ Hits                   38849    38854    +5     
  Misses                 14262    14262           
+ Partials                5981     5980    -1     

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

[GitMensch]

Is there a reason to skip the complete termination?
I think we should do the screenio cleanup - but only if that was done after the call to cob_call_with_exception_check().
We very likely need to call the exit handlers, as those can be COBOL programs which need the state "at exit", not any previous state we longjmp to. But then those should not be executed after the longjmp.

We possibly should add some additional testcases that combine cob_call_with_exception_check() with the scenario "exception and exit handlers setup".

[ddeclerck]

Is there a reason to skip the complete termination?

Well, in #125, you suggest :

I'd prefer the second - and document that modules will only be unloaded with this function if after the call a manual call to cob_tidy() is done.

Since cob_tidy calls call_exit_handlers_and_terminate, it seemed natural to just skip the whole termination.

Anyways, I pushed an improved version, where only the last bits of termination are delayed, ie only the following calls :

	cob_exit_call ();
	cob_exit_common ();

Cleanup is automatically performed after the longjump, so no need to call cob_tidy.

[GitMensch]

That's very clean. Please add a changelog entry, then this is good for svn (sight, I won't finish the parts to make svn ready for more commits until moid/end next week).