Quality improvment of Similar question Feature

## Is your enhancement request related to a problem? Please describe
Right now it looks query condition is like '%Input_string%' this is only good first to start.
I have testing this on meta.answer.dev

## Describe the solution you'd like
Use tokenizer (quite common in Nlp) which split input text into small words (sometimes sub words) and lookup tokens in database for finding similar question

## Describe alternatives you've considered

A clear and concise description of any alternative solutions or features you've considered.

No files or tests are named. Start by locating the similar-question implementation and its current database query, then study how question text is stored and searched. Done means a defined, reviewed tokenizer-based matching approach with evidence that it improves similar-question results over the current substring lookup.

Written by the indexing model from the issue text.