worker: report cashes directly to logrus

This is a bit of an RFC commit, I noticed that when we discussed
a crash from the worker we looked at individual message from
syslog/journald for the stacktrace deatils. I was wondering if
having a more direct crash report would be more useful? We can
of course also add more logrus features to flag those with tags
like "crash" or something (I did not do that in this PR, I don't
know much about the operational side, sorry).

cmd/osbuild-worker/main_test.go

@@ -0,0 +1,41 @@
+package main_test
+
+import (
+	"io"
+	"testing"
+
+	"github.com/sirupsen/logrus"
+	logrusTest "github.com/sirupsen/logrus/hooks/test"
+	"github.com/stretchr/testify/assert"
+
+	main "github.com/osbuild/osbuild-composer/cmd/osbuild-worker"
+)
+
+func TestCatchesPanic(t *testing.T) {
+	restore := main.MockRun(func() {
+		// simulate a crash in the main code
+		var foo *int
+		println(*foo)
+	})
+	defer restore()
+
+	// logrus setup is a bit cumbersome as we need to modify both
+	// the standard logger and add a mock logger.
+	var exitCalls []int
+	logrus.StandardLogger().ExitFunc = func(exitCode int) {
+		exitCalls = append(exitCalls, exitCode)
+	}
+	logrus.SetOutput(io.Discard)
+	_, hook := logrusTest.NewNullLogger()
+	logrus.AddHook(hook)
+
+	main.Main()
+	// ensure both message and stracktrace are reported in full
+	assert.Equal(t, logrus.FatalLevel, hook.LastEntry().Level)
+	msg := hook.LastEntry().Message
+	assert.Contains(t, msg, "worker crashed: runtime error: invalid memory address or nil pointer dereference")
+	assert.Contains(t, msg, "runtime/debug.Stack()")
+	assert.Contains(t, msg, "osbuild-worker_test.TestCatchesPanic.func1()")
+
+	assert.Equal(t, []int{1}, exitCalls)
+}