fix some memory leaks

[dirk-thomas]

This patch fixes some of the memory leaks seen when running the demo_nodes_cpp talker / listener.

But there seem to be several more - mostly inside OpenSplice, e.g. the following - maybe @michielb-prismtech has an idea if we are using the API wrong anywhere:

The call to tsMetaHolder->create_datawriter (https://github.com/osrf/opensplice/blob/0941db497e2952ddf69454442071e92e0e69d942/src/api/dcps/c%2B%2B/common/code/Publisher.cpp#L257) in Publisher::create_datawriter allocates memory.

But the corresponding call to deinit (https://github.com/osrf/opensplice/blob/0941db497e2952ddf69454442071e92e0e69d942/src/api/dcps/c%2B%2B/common/code/Publisher.cpp#L375) in Publisher::delete_datawriter seems to never deallocate the memory hence the leak shown in valgrind:

==7156== 352 (248 direct, 104 indirect) bytes in 1 blocks are definitely lost in loss record 402 of 487
==7156==    at 0x4C2E0EF: operator new(unsigned long) (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==7156==    by 0xAABBCB4: rcl_interfaces::msg::dds_::ParameterEvent_TypeSupportMetaHolder::create_datawriter() (ParameterEvent_Dcps_impl.cpp:45)
==7156==    by 0x9023E83: DDS::OpenSplice::Publisher::create_datawriter(DDS::Topic*, DDS::DataWriterQos const&, DDS::DataWriterListener*, unsigned int) (Publisher.cpp:257)
==7156==    by 0x821FDBC: rmw_create_publisher (rmw_publisher.cpp:160)
...

[mikaelarguedas]

🎉 👍