feat: handling synonyms and text fields more efficiently

[alexgarel]

• Using synonyms capabilities of ES to avoid storing taxonomies fields in the index
• Better handling of full text queries that support them within any expression
• make boost_phrase a separate parameter
• Raise errors if the query is not well understood or do not pass some sanity checks
• Use main translation of taxonomy for facets values (instead of a random synonym)
• Better handling of global config to avoid treacherous patterns
• Unify parameters for Get and Post (better use of pydantic)
• Error on extraneous search parameters to avoid hard to debug issues with typos

Part of: #193

@raphael0202 revealed some pending issues:

• make labels:"fair-trade" work (it does not work because of the "-" being splet in two words)
• make labels:"appellation d'origine protégée" work (it does not work because of apostrophe

It might be a change in analyzer and/or synonyms file.
Apart from adding new test in test_search.py, I need to add a test on ES analysis, as it will help document behaviour.

[raphael0202]

We name the logger logger in the rest of the project, and use app.utils.get_logger function (to set-up the right LOG_LEVEL)

[alexgarel]

I did this because the logging.getLogger is very strange… if you give it a name, it resets log level to NOT_SET !

Here it wasn't used anymore so I removed it.

I will do a refactor on that, some time ;-)

[raphael0202]

Great work! 🎉
I tested it locally, it works well :)