Allow configuring fallback and/or preferred langid for USB strings

[jimklimov]

Just so we do not forget your earlier comment: the langid number (and/or name for reporting?) might be better configurable by end-users.

Low priority, gotta see if devices exist that could benefir from this at all.

Originally posted by @jimklimov in #2604 (comment)

See also: #2604 (comment) about use of existing *langid* options.

[jimklimov]

Per subsequent discussions in other issues, there is also a possibility that a device would serve several locales for some or all strings. A program is then free to extract each data point in desired locale (may be different locales for different queries), typically this happens in user-interfacing programs based on their run-time choice of locale. This may apply to NUT drivers as well, although it feels better to let the user prefer a locale (list?) explicitly.