Send the lh update worker into its own routine instead of taking over the reload routine

[nbrownus]

@johnmaguire came across this issue today.

If the lighthouse.interval config is changed and nebula reloads it then the reload routine is taken over. This will cause nebula to be unable to shutdown cleanly and block future reloads from occurring.

goroutine 11 [select]:
runtime.gopark(0xc0001f56b0?, 0x2?, 0x0?, 0x0?, 0xc0001f567c?)
        runtime/proc.go:381 +0xd6 fp=0xc0001f54f8 sp=0xc0001f54d8 pc=0x43b9b6
runtime.selectgo(0xc0001f56b0, 0xc0001f5678, 0xc000191200?, 0x0, 0x0?, 0x1)
        runtime/select.go:327 +0x7be fp=0xc0001f5638 sp=0xc0001f54f8 pc=0x44bc5e
github.com/slackhq/nebula.(*LightHouse).LhUpdateWorker(0xc000184a50, {0xd3e608?, 0xc000036fa0}, {0xd3e8e0?, 0xc000191200})
        github.com/slackhq/[email protected]/lighthouse.go:774 +0x20c fp=0xc0001f56e8 sp=0xc0001f5638 pc=0xa70dac
github.com/slackhq/nebula.(*LightHouse).reload(0xc000184a50, 0x1171f00?, 0x0)
        github.com/slackhq/[email protected]/lighthouse.go:220 +0x775 fp=0xc0001f5b18 sp=0xc0001f56e8 pc=0xa6c315
github.com/slackhq/nebula.NewLightHouseFromConfig.func1(0xc000034cc0?)
        github.com/slackhq/[email protected]/lighthouse.go:134 +0x32 fp=0xc0001f5ba0 sp=0xc0001f5b18 pc=0xa6b9d2
github.com/slackhq/nebula/config.(*C).ReloadConfig(0xc000034cc0)
        github.com/slackhq/[email protected]/config/config.go:158 +0x3a3 fp=0xc0001f5e18 sp=0xc0001f5ba0 pc=0x896383