I am sorry that I can not give you a well researched source, in fact I would be interested in one myself.
If you are concerned about data collection, you probably should not use any modern smart device. Be it from Google, Apple or Microsoft. Companies earn huge amounts of money by collecting data. You can disable a lot of features and set app permissions, but you will also limit functionality doing that. And additionally you don't know if the settings really disable anything.
You can read the privacy policies, but I doubt that will tell you anything.

Holy shit you're paranoid. I hope you don't have a phone because that is 10 times worse. Google doesn't "spy" on you. They are 100% open about what they have collected about you, you can view it any time and you can take all your data and leave if you want. You really need to educate yourself better. Stop listening to the technopanic

That said....

• What data is collected by Google/Nvidia/tiers companies, when I use the Shield? why and how?

Google will track anything tied to your account. If you use Google to log in to something, they will have a record of it and what you did. Everything on YouTube or any other Google app. Other apps will track your usage of those services.

• Can I disable the "always listening" function? (really scary) I don't really need voice command, if needed I mostly prefer talking to the remote after pressing the mic button.

Yeah... Totally scary! Wait.. No, it's not. That said, the OG shield doesn't have any always listening. There is a button you press that activates the mic. It should be noted, no device with a microphone is safe, they ALL can be remotely activated without notification. If you're that paranoid, don't have any device with a mic. Period.

The new Shield has always listening but no one has it yet so we don't know if can be turned off. Common sense says it will be. But refer above.

• Should I be worried about anything else? I'm no privacy freak, it's just a TV box, but I don't want to be "fully open" to Google either.

Well, the EULA requires you hand over your first born child. Personally I'm ok with it because he's a bit of an asshole and the Shield is just that good .

• Also a more practical question : I don't have any Android device, will it impair my use of the ShieldTV? I have a Youtube/Gmail account, I guess it's ok.

Nah. The ShieldTV operates independently from a phone and it's not particularly "Google heavy" if you don't want to. The non-google services get equal footing as any others, use em or don't.

• If I ever get an Android smart TV (or any other TV OS) as it is becoming the norm, how will it interact with the ShieldTV? Will I have a hard time with duplicate functions?

That depends entirely on you. Any smart TV will duplicate most functions of the shield. The shield will do it better 99% of the time, so most often you'll be using one or the other.

Wow people in this sub are dicks. OP cares about privacy and everyone freaks out. Stay classy.

Don't have any electronics if you're that paranoid.

I'm not THAT paranoid, I'm concerned, skeptical. I know I own technology that records usage data (Skype, Evernote, Facebook, Windows 10 for instance, even if I removed all the options that sounded bad. And I think that if Android had horrible spying habits, it wouldn't be that widely used.

The thing is, I'm ready to accept a little "spying" for great advanced functionalities and user experience that I can't provide myself. A good example being voice command : I'm ok if it sends [b]text-translated requests[/b] in relation to my usage, I'm not if it's constantly recording and transmitting what happens in the room. Cause I probably can DIY a Raspberry Pi with open-source basic recognition software who will perform less good, but don't feed on my data.

That's why I don't own an Android/iOS smartphone yet : I'm not that concerned, but the apps and features aren't that better/useful for me compared to what my Blackberry does. Not useful enough compared to my concerns.

On an AndroidTV device, I'm ok if it's gathering data about my usage of said device (media consumption, Youtube, Netflix, Plex...). I'm not if it's analysing my every move and voice command. But giving up my firstborn is ok. He'll probably be a ginger anyway.

Also, how do they make money from my data? Ads? I don't see ads. For Apple/Microsoft I guess is by selling high-end hardware, but Google? They make free services, and win that much on my data?