Voynichese is a numeric cipher?

[Jorge_Stolfi]

this „weirdo“ (Stolfi‘s branding)

It is not my branding.  Back in the mailing list days, "weirdo" was standard Voynichology nomenclature for glyphs that occurred only once or twice in the whole book.  Is that term taboo now? Was Beinecke sued for offensive language by the International Association of Special Glyphs and Symbols?

the right part of „the red“ is very weak

That part lies withing a stain which looks like a liquid spill.  I would think that its red paint and outline were originally as strong as those in the left part, but were mostly erased by the liquid.

the outlined parts and stronger red colour may have been [...] also contact-transferred by closing the books‘ pages a bit too early.

I don't think it is a likely scenario. The paint is too strong, and the outline is too sharp and continuous. An offset print would be even fainter and more irregular than the right part.

Besides, even flipped left-to-right it does not look like anything. 

About VMS being a numerical cipher: what would that change or deliver?

If it is a codebook cipher, it will be harder to crack than any letter-based algorithmic cipher. But not impossible. However, the solution would come come out piecemeal, one word at a time, rather than in one finger snap.  Like the solution of Etruscan, or of the Rohonc codex.

Character statistics would be useless.  Instead one would have to guess the meaning of certain key words and phrases.  Hopefully the Herbal section will contain recurring phrases like "make into a tea" or "will be cured" or "it is proven", which could provide useful "cribs".

One may begin to solve it even without identifying the language.

All the best, --stolfi