nut-scanner: fix device-name collisions in generated conf

[aquette]

When nut-scanner is called separately, to scan different buses, it will generate similar names for each of these:

• USB scan

[nutdev1]
...

• NUT scan

[nutdev1]
...

• ... As a result, one can't just redirect nut-scanner outputs directly to ups.conf, without a 'sed' hack workaround.

Solution: simply refine the device names, like "nutdev-".
Ex: nutdev-usb1, nutdev-usb2, nutdev-nut1, nutdev-snmp1, nutdev-simulation1, ...

Closes: #2243

UPDATE/FYI: As recognized during investigation for #2317 the actual device numbering (in case of multi-bus scanning) is still monotonous, assigned during ups.conf style rendition of discovered configuration (actually not defined outside a renderer). So a single run with many media types enabled would yield e.g. "nutdev-usb1, nutdev-usb2, nutdev-nut3, nutdev-snmp4, nutdev-simulation5, ..."

[aquette]

@jimklimov CI issues, unrelated to the PR (trivial) changes
cc @desertwitch for review (interested in an onboarding?)

[desertwitch]

Currently have no means to test (stuck on a train) but the principal idea is great and should make chaining different nut-scanner commands, while outputting into one configuration file, much easier and reduce the need to hack around the generated UPS names. This'd also be useful for my use cases.

[aquette]

Note that #2246, as for all my dummy-ups work, is there for when you don't have access to real devices (or for CI). I coded that when on a plane ...
Just touch some .dev files in /etc/nut and try nut-scanner -n

[jimklimov]

For coding on a plane or a boat, did I already suggest Termux? An Android app that is a debian-ish Linux VM, can build NUT for native ARM for example :)

[aquette]

Got it on all phones and tabs 😉
But home assistant on my rpi gives me remotes access to my editable containers and has a term and file edit through web UI too, quite nice. Give it a try ...

[jimklimov]

Finally back to the online world, re-read this PR and was like "how did I never hit this, aren't numbers sequential for all found devices on whatever bus?"... and then saw the part about "running separately for different buses". Makes sense and LGTM then :)

[AppVeyorBot]

❌ Build nut 2.8.1.1274-master failed (commit b11207fd48 by @jimklimov)

[jimklimov]

@aquette : changes seem to work, NIT complains about test result mismatch (will post a fix in a sec) ;)

Now that we have a ./ci_build.sh here, feel free to use it locally to weed out at least simple hiccups without incurring the wrath of the whole CI farm :)