ko run does not delete pod upon interrupt

Hitting Ctrl+C after pod has started leaks the pod:

```
Flag --generator has been deprecated, has no effect and will be removed in the future.
If you don't see a command prompt, try pressing enter.
<>

^C2021/02/03 10:46:55 error executing "kubectl run": signal: interrupt
```

```
NAME READY STATUS RESTARTS AGE
cancel-10aec35353f9c4096a71c38654c3d402 1/1 Running 0 57s
```

from the speed of ctrl+c it seems it's not even calling into kubectl.

I compiled from HEAD yesterday. 82cabb40bae57

Start by reproducing the issue with the ko run command and interrupting it after the pod starts. Trace the command's interrupt handling around kubectl run; done means Ctrl+C exits without leaving the running pod shown in the report.

Written by the indexing model from the issue text.