feat: support nbest output in normalize/tag

[pengzhendong]

Closes #296

Summary

Add nbest parameter to normalize() and tag():

• normalize(input) — returns a single string (default, backward compatible)
• normalize(input, nbest=N) — returns a list of N-best results

Examples

from tn.chinese.normalizer import Normalizer
n = Normalizer()

n.normalize('海淀区108号', nbest=3)
# ['海淀区一百零八号', '海淀区十八号', '海淀区幺零八号']

n.normalize('收费24.0', nbest=3)
# ['收费二十四点零', '收费二四点零', '收费2四点零']

Test plan

• All 1400 unit tests pass (backward compatible, default nbest=1)
• CI passes