This fixes several raw text related problems for new sentences and reports.

• Introduces simple central functions
• All free-form text input (sentences, sentence sources, text from reports' other option) will be cleaned before entering the database.
• The sentence_id is calculated from the cleaned sentence

Notes/warnings:

• It does not handle any existing data. We still need cleaning in bundler for metadata .tsv files so they have correct structure (columns) coming from past malformed data.
• This PR fixes [BUG] Irregular spacing between words can result in duplicate sentences added via write page #4614 but if there already is a sentence entered with double-space (or TAB, CR, LF, ...) the new & cleaned sentence will be allowed.
• In dev environments, where sentences are imported from files, there might be cases where the sentence_ids / even sentences not match, because this time they will be cleaned of invisible characters like TAB or CR/LF (word separating characters).
• Some existing regex cleaning has been replaced by this, but the piped one (server/src/core/sentences/cleaning/multiple-sentences.ts) is untouched.

Limited testing on dev environment - checked from local database (Dockerized):

• Entering a single sentence with TAB chars in sentence and source => OK
• Entering a small batch sentences with (malformed sentence and source) => OK
• Reporting with "other", multiple lines and TAB characters => OK