fix: add tokenizer for the query string

[BlessSan]

Problem: lunr query search term depends on spaces for each token. However, lunr.ja plugin tokenize the fields (documents) using japanese segmenter.

So, for example the resulting token created for the field 標準製品仕様は will be 標準, 製品, 仕様, and は (as demonstraded in http://chasen.org/~taku/software/TinySegmenter/)

Solution: To search for 標準製品 (which is two words), tokenize the input query using the same tokenizer. Then convert it back into lunr query search term.

Limitation: since this method essentially add spaces to Japanese words (as dictated by the segmenter used by lunr.ja). Using wildcards, fuzzy, boosts (https://lunrjs.com/guides/searching.html) in search method would produce false result.