cmd/utils: don't enumerate usb when --usb isn't set

[gballet]

USB enumeration still occured. Make sure it will only occur if --usb is set.

[fjl]

I think this is not right and it will break the config file.
What needs to happen is: the default of NoUSB in the struct needs to be made true.
Then, if the flag is set, it needs to be changed to false.

If someone has NoUSB = false in the config file, it should still work.

[gballet]

Since it doesn't seem to set default values in the deserializer, the default has to be set either before or after deserializing. How do you propose to distinguish the case when NoUSB is absent from the TOML file, from the case when the user explicitly set NoUSB = false ?

[gballet]

It is very unlikely that people will want to set NoUSB = false because until now, NoUSB = false was the default and now if they want to enable USB by default they don't have an option for that yet. What I suggest is also add a USB command in the config that will default to false, and that people can set to true.

[gballet]

@fjl I iintroduced a config.USB option to remove the ambiguity introduced by the previous PR. If people set config.NoUSB in their TOML file, then it will display a warning. If people want to enable USB by default, they can either use --usb or set USB = true in their TOML config file.

[renaynay]

LGTM

[holiman]

I think there are a few more places you could clean up.

[user@work go-ethereum]$ grep -r "nousb" .
./cmd/geth/les_test.go:	args = append([]string{"--networkid=42", "--port=0", "--nousb", "--ipcpath", ipcName}, args...)
./cmd/geth/les_test.go:	args := []string{"--nousb", "--networkid=42", "init", "./testdata/clique.json"}
...

[holiman]

LGTM