OS signals may be received by background thread

[jcheng5]

(We don't know for a fact that this is happening, it's just a theory)

The libuv io thread isn't masking SIGINT (or any other signals), it's conceivable that a SIGINT sent to the R process could be handled by our thread instead of the main R thread. We are hypothesizing this because @nstrayer saw cases where SIGINT did not cause an R process to exit.

[wch]

This says (https://stackoverflow.com/a/22005820/412655):

If you send a signal to a process, which thread in the process will handle this signal is undetermined.

We could test it by adding a SIGINT handler on the background thread and have it print out (or log to disk) whenever it receives a SIGINT.