Fix SIGSEGV in kubernetes executor Cleanup

Kubernetes executor Cleanup() assumes POD has been created, but it will be invoked also in case of failure during POD creation.

Closes #2553 (closed)

    Starting Kubernetes command...                      job=43034187 project=1208199 runner=2adc161b
    ERROR: Job failed (system failure): Post https://192.168.64.4:8443/api/v1/namespaces/default/secrets: x509: cert
    Submitting job to coordinator... ok                 job=43034187 runner=2adc161b
    panic: runtime error: invalid memory address or nil pointer dereference
    [signal SIGSEGV: segmentation violation code=0x1 addr=0x48 pc=0x1fd1e23]

    goroutine 55 [running]:
    gitlab.com/gitlab-org/gitlab-runner/executors/kubernetes.(*executor).Cleanup(0xc4203de000)
            /Users/nolith/src/gitlab.com/gitlab-org/gitlab-runner/executors/kubernetes/executor_kubernetes.go:164 +0
    gitlab.com/gitlab-org/gitlab-runner/common.(*Build).Run.func1(0xc42027fb80, 0xc420402dc0, 0x3f59da0, 0xc42009e28
            /Users/nolith/src/gitlab.com/gitlab-org/gitlab-runner/common/build.go:372 +0x189
    gitlab.com/gitlab-org/gitlab-runner/common.(*Build).Run(0xc420402dc0, 0xc42008ad00, 0x3f59da0, 0xc42009e280, 0x3
            /Users/nolith/src/gitlab.com/gitlab-org/gitlab-runner/common/build.go:403 +0x5ea
    gitlab.com/gitlab-org/gitlab-runner/commands.(*RunCommand).processRunner(0xc4202a9700, 0x0, 0xc4204511e0, 0xc420
            /Users/nolith/src/gitlab.com/gitlab-org/gitlab-runner/commands/multi.go:175 +0x7c3
    gitlab.com/gitlab-org/gitlab-runner/commands.(*RunCommand).processRunners(0xc4202a9700, 0x0, 0xc42006da40, 0xc42
            /Users/nolith/src/gitlab.com/gitlab-org/gitlab-runner/commands/multi.go:183 +0x211
    created by gitlab.com/gitlab-org/gitlab-runner/commands.(*RunCommand).startWorkers
            /Users/nolith/src/gitlab.com/gitlab-org/gitlab-runner/commands/multi.go:199 +0x9f