ole32/tests: guard ReleaseStgMedium with SUCCEEDED

[Joerg-Cyril.Hoehle at t-systems.com]

Dear reviewers,

Please assess this code carefully (as always), as the clipboard resource
allocation protocol is unknown to me and Alexandre Julliard already
spotted one error there (corrected herein).
The only things I'm sure about are:
- the bogus Release in line 480;
- sometimes releasing sometimes not is bogus (around line 530);
- the unconditional Release and crash in line 1358 when the prior call fails.
The rest (unconditional Release or execution) just did not look right to me.

This prevents the crashes mentioned in bug #20204 and test.winehq.
It does not explain why clipboard metafile retrieval fails on 
some machines.


Regards,
	Jörg Höhle.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: wine-ReleaseStgMedium.patch
Type: application/octet-stream
Size: 9330 bytes
Desc: wine-ReleaseStgMedium.patch
URL: <http://www.winehq.org/pipermail/wine-patches/attachments/20090929/329ee993/attachment-0001.obj>