Voynichese is a numeric cipher?

[ololololo]

However, I don't consider this a reason to criticize the theory of numerical encryption in general, as it aims to try to reproduce the VMS text rather than to find out how quickly I can solve anagrams.

I'm not sure one needs a reason to criticize various aspects of different theories. For example, it's clear from this experiment that the numerical encryption is not very convenient. And even the author of the solution cannot manually decode it, so I think it's safe to say it may not be a very practical cipher. It's also not clear whether this cipher can reproduce the statistics of the Voynich MS. Even with a large number of nulls one has to propose a way to place these nulls that would produce a result similar to the Voynich MS. I don't think it's enough to say "in principle nulls can reproduce any distribution, so this explains all the statistics". I think it has to be shown that some specific way of placing nulls, randomly or pseudorandomly or according to some simple rules, will produce a result similar to the Voynich MS. In general, there should be an explicit algorithm, or better a piece of code, that receives some plaintext and produces a good likeness of Voynichese to prove that a cipher actually a good candidate.

I totally agree with you!
It is not difficult to create rules for nulls. For example, ch and yk/yt will most often be placed before the digrams ol/or and eor/eol/ey.
Final units can be converted to y or m.
The final e in the words can be followed by a y.
There is a possibility that Currier A has something similar.