User-space stack backtrace

This was kind of discussed in [issue #1070](https://github.com/iovisor/bcc/issues/1070) but I'm having trouble getting user-space backtraces from 32-bit ARM binaries. If I take a simple test application I get proper user space callchains for the 64-bit binary but not the 32-bit binary. For the 32-bit binary I just get one instruction pointer/program counter so it seems like the in-kernel unwinder has trouble. The unwind information is stored a bit differently in 32-bit ARM binaries so I wonder if that's the problem. Are there any workarounds? @4ast mentioned possibly reading some of the stack memory and unwinding in userspace like perf allows you to do. Are there any examples of that in bcc?

No source files, tests, or entry points are named. Start by reading issue #1070 and comparing the reported 64-bit and 32-bit ARM backtrace behavior; done would mean demonstrating a reliable multi-frame user-space backtrace or documenting a verified workaround and relevant BCC example.

Written by the indexing model from the issue text.