WIP Print backtrace on failed ASSERT_PKRUs

[ayrtonm]

This currently prints a trace but at least brief checks on the read_config test shows the trace ends at compartment boundaries. Need to fix that, add anything else that might be useful and swap IA2_DEBUG for IA2_RELEASE with opposite semantics. Closes #452.

[ayrtonm]

I think the issue with the shortened backtraces is specifically in IA2_CALL callgates, not general cross-compartment calls.

./tests/read_config/read_config 
#0 0x70cc4a79fdd6 in /home/ayrton/ia2/IA2-Phase2/build/tests/read_config/libread_config_call_gates.so(ia2_print_backtrace+0x16) [0x70cc4a79fdd6] ()
#1 0x70cc4a79e86a in /home/ayrton/ia2/IA2-Phase2/build/tests/read_config/libread_config_call_gates.so(__wrap_parse_str+0x36) [0x70cc4a79e86a] ()

the trace ends at the PLT stub so the issue might not be related to the call gate prologues.

[ayrtonm]

Printing the backtrace for the broken test mentioned above gives this

0x00007ffff7e0b874 in __wrap_parse_str () from /home/ayrton/ia2/IA2-Phase2/build/tests/read_config/libread_config_call_gates.so
(gdb) bt
#0  0x00007ffff7e0b874 in __wrap_parse_str () from /home/ayrton/ia2/IA2-Phase2/build/tests/read_config/libread_config_call_gates.so
#1  0x00007ffff7e1459c in __ia2_parse_str () from /home/ayrton/ia2/IA2-Phase2/build/tests/read_config/libread_config_lib.so
Backtrace stopped: previous frame inner to this frame (corrupt stack?)

Although call gate control flow is wrong here (__ia2_foo should call foo not __wrap_foo) this is exactly the type of case that we want to be able to handle so we should do whatever gdb is doing.