you suggesting that gender in these languages was an intentional decision to solve the problems you raise?

No, I'm not suggesting intention or decision. Most of the time, language works a lot like a biological species: there's no critter or speaker deciding "we shall have this feature!", but instead the feature spreads or goes extinct depending on the role that it performs in the language, alongside other features.

My explanation is all about that role. That is the point of grammatical gender, and it explains:

• why it appeared independently across different languages? Clearly the gender systems in Dyirbal, in most Indo-European languages, in [most?] Afro-Asiatic languages are unrelated to each other, but why did they develop that same feature?
• why it survives for so long in a language? For example, the gender system in Russian, Hindi and Spanish backtracks all the way back into Late Proto-Indo-European (6000? years ago).

A pointless feature wouldn't do it.

I do not remotely know Portuguese, but how does this derivation quality help with the word for an apple seed?

The fruit vs. plant example is from Italian, not Portuguese (see note*).

It doesn't need to help with the word for an apple seed (IT: seme di mela, lit. "seed of apple"). It's just an extension, a "bonus" of the system; the core is like bambino/bambina, words referring to human beings, we humans tend to speak a lot about each other.

That said, your question reminds me the noun classes of Bantu languages. Gender is just a specific type of noun class; it's possible that some language out there would actually use a noun class derivation of their word for apple to refer to apple seeds.

Languages involved naturally. Nothing is an intentional decision.

Yes, it's still possible to have ambiguity, just like you can have hash collisions in hash tables. But it at least sometimes helps.