Commands timing out produces very ugly and inconsistent error messages

I get:
```
Error: Post http://127.0.0.1:5001/api/v0/name/publish?arg=QmShyawfbYCT6ZMDawg8C8dmhSh4UWXGFfjHvqMfB8npny&encoding=json&key=self&lifetime=24h&resolve=true&stream-channels=true&timeout=1s: context deadline exceeded
```

Sometimes, and other times just:
```
Error: context deadline exceeded
```

Both of these are ugly (should just say command timed out) or something. I think the first one is caused by the server timing out and closing the connection, and the second one is the client beating the server to the punch.

cc @keks

Reproduce the timeout against the CLI and the /api/v0/name/publish endpoint, checking both the server-closing-connection case and the client deadline case. Trace where each error reaches the command output, then verify that both paths produce one consistent, user-friendly timeout message.

Written by the indexing model from the issue text.