Worker initialization sometimes times out while accessing IAS

Sometimes the worker host aborts the IAS /report request. It respawns the worker and can recover though.

Details

looks like this:

level=error module=grpc-9988 caller=grpc.go:169 ts=2019-01-08T02:50:03.878308306Z msg="request failed" method=/ias.IAS/VerifyEvidence req_seq=30 err="ias: http POST failed: context deadline exceeded"
level=error module=worker/host/sandboxed caller=protocol.go:191 ts=2019-01-08T02:50:03.949256085Z runtime_id=0000000000000000000000000000000000000000000000000000000000000000 worker_pid=17995 msg="error while receiving message from worker" err=EOF
level=warn module=worker/host/sandboxed caller=sandboxed.go:98 ts=2019-01-08T02:50:03.985679123Z runtime_id=0000000000000000000000000000000000000000000000000000000000000000 worker_pid=17995 msg="worker process terminated"
level=error module=worker/host/sandboxed caller=protocol.go:41 ts=2019-01-08T02:50:03.985722293Z runtime_id=0000000000000000000000000000000000000000000000000000000000000000 worker_pid=17995 msg="error while closing connection" err="close unix /tmp/ekiden-worker460609677/host.sock->@: use of closed network connection"
level=error module=worker/host/sandboxed caller=sandboxed.go:701 ts=2019-01-08T02:50:03.985737178Z runtime_id=0000000000000000000000000000000000000000000000000000000000000000 msg="failed to spawn new worker" err="worker: error initializing SGX CapabilityTEE: worker: error while verifying attestation evidence: rpc error: code = DeadlineExceeded desc = context deadline exceeded"

cross-reference: #1262

Acceptance Criteria

• that it times out very rarely

Start with grpc.go:169, protocol.go:41 and :191, and sandboxed.go:98 and :701, then review cross-reference #1262. Trace the IAS /report or VerifyEvidence deadline through worker initialization and use the supplied logs as the failure case. Done means the worker initialization timeout occurs only very rarely, as specified by the acceptance criterion.