Investigate whether FAF maps addresses in stack traces to correct line numbers

[mmilata]

On x86 stack, return addresses are stored on the stack that point to the next instruction (which debuginfo might translate to next line). Do we take this into account?

Example:
http://mmilata.fedorapeople.org/firehose-crash-correlation/rhytmbox/sources/e14992209f19440e0304e736d14d7e9d1c94dd67.html#file-e14992209f19440e0304e736d14d7e9d1c94dd67-line-1626
https://retrace.fedoraproject.org/faf/reports/21759/

rhythmdb_query_model_do_insert calls rhythmdb_query_model_insert_into_main_list on line 1626 but FAF says it does so on 1629 (line with next statement).