Introduce Crash Monitoring

[fxbox]

Introduce Crash Monitoring

[Yoric]

Apparently, we should be able to reuse the Crash Monitor.

Filing a bug to be able to upload data to Socorro.

[Yoric]

Extracts from irc:

7:10:21 PM - ted: https://chromium.googlesource.com/breakpad/breakpad/+/master/docs/linux_starter_guide.md
7:10:32 PM - ted: Yoric: if you're only targeting linux then it's pretty simple
7:11:07 PM - ted: use ExceptionHandler to generate dumps, do something in the callback to invoke another process to submit them (and maybe restart yourself)
7:11:26 PM - ted: then also run dump_syms on your binaries as part of your build and upload the results to the symbol server
7:22:35 PM - ted: well i don't really have an answer for rust

7:22:52 PM - ted: but you could write a little c++ file that did that and exposed a tiny C API around it
7:23:08 PM - ted: foo* set_exception_handler()
7:23:17 PM - ted: unset_exception_handler(foo*)
7:23:40 PM - ted: i feel like someone in #rust or #servo was looking into integrating breakpad
7:23:47 PM - ted: but i don't know where that got to
7:24:09 PM - ted: also the more basic question of what crash reporting means in rust code is not something i understand terribly well, tbh
7:24:21 PM - ted: handling OS-level crashes still makes sense
7:24:29 PM - ted: but dealing with unhandled panic! etc

[samgiles]

We'd need some sort of process manager/API within the FoxBox process too.

A large part of the tunnel_controller is managing the process running the HTTP tunnel to the cloud service. I'm wondering if there's some overlap here - will each device_adapter be a process as well?

[Yoric]

will each device_adapter be a process as well?

This would make sense. It would make it easier to add/upgrade drivers without having to reboot the entire FoxBox feature.

[Yoric]

Filing a bug to be able to upload data to Socorro.

Ok, we apparently don't need any authorization to upload to Socorro. We just need to decide (internally) of a product id.