Docs caught lying about dependency resolution

[2colours]

As per this commit, there is a part on the https://docs.raku.org/language/modules page that reads like (with some additional annotations from me)

It should also be noted that when writing the depends section of the META6.json file, it is the name of the distribution that is listed, not the name of the compunit that is desired[1]. Although, for very many entities, the name of the distribution and the name of the compunit are the same. Another global effect of the depends list in a distribution made available to the Ecosystem, is that package managers, e.g. zef, can look for compunit names in all distributions available[2].

[1] is simply not true. I don't know if it was true with p6c and some obsolete package manager but it's clearly not true for a long while, and it's also documented (with the example I'm going to use) that Richard didn't know that at the time of writing this bit.
If we take Pod::Render from MARTIMM and try to install it using zef install Pod::Render, it will not only fail but it can be noted that it tried to install an old version of dist Raku::Pod::Render (more precisely: Raku::Pod::Render:ver<3.5.2>:auth<github:finanalyst>), rather than the dist literally called Pod::To::HTML, as it was originally intended.

[2] is strange here - if we think it further, it could lead us to the right conclusion that contradicts point [1]. If the name of the distribution would be used in "depends", why would it matter that package managers "can look for compunit names [inside]"? This is a dangling statement. Anyway, actually, they (or zef at least - I don't know if this is specified behavior!) seem to look for compunits (modules? - whatever, I'm always confused about this anyway), and clearly not distributions, so there is that.

[2colours]

I tried to see how zef actually does dependency resolution but the code is fairly convoluted for me still...