Provide quantity support for Serbian

[nciric]

Problem:
Demo program at https://github.com/nciric/test-inflection/blob/main/quantity.cc outputs the following for two words (one in lexicon and the other one not).

Run it with a word from a dictionary, like девојка, the output is
print: 1 девојка
speak: једна девојка
print: 2 девојке
speak: две девојке
print: 1 девојци
speak: једна девојци (NOTE: wrong case for number, RBNF update needed)

Run it with a word outside dictionary (rule based), like боја, the output is
print: 1 боја
speak: 1 боја
print: 2 боја
speak: 2 боја
print: 1 боји
speak: 1 боји

It's worse in two aspects:

1. Speak doesn't "kick in" as I always get 1 or 2 as digits
2. It seems not to inflect (except for the last 2, which is even more confusing)

Two things have to happen to fix this issue:

1. ICU needs updated RBNF forms
2. Common concept factory needs to be updated, e.g. here's Russian example inflection/src/inflection/dialog/language/RuCommonConceptFactory.cpp