One scar away from losing access to your ability to pay …
Biometrics can not really be changed. Except maybe through time or trauma (i.e. age or injury). They can be used to uniquely(?) identify a person - except maybe twins - at the expense of anonymity, which has it's own set of problems.
But because they can not easily be changed they're a terrible security feature. Once they leak, they're unusable and you're hosed. You can't issue a new palm print for your bank account like you could a new chip card and password.
Also, just because you waved your hand over a scanner does not mean that you approve and consent of the transaction. With tap to pay there were ideas of mobile point of sales devices just tapping on peoples backpacks in a crowded area. You don't even keep your biometrics markers in your pocket, they're just out in the open for anyone with a camera. This may be bordering on paranoia, but a few years back (2014) German hackers from Chaos Computer Club took iris scans from Angela Merkel (then Chancellor of Germany) and finger prints of Ursula von der Leyen (then Minister of defense) using nothing but press fotos. Cameras have only gotten better.
TL;DR: Biometrics can be used for identification but should never be used for authorisation.
You should absolutely have web environment integrity. Your browser should not allow the website to do things that you don't approve of, so the integrity of your computer can be ensured.
Wait, that's not what they mean, is it? Oh no … 🙄
Yea, I feel like Google has this a bit backwards. As always, I like to turn the metaphor on it's head. You're not visiting a website, you're inviting a website. You're allowing the website to use your system resources, bandwidth, CPU cycles, etc. And what you do with your own system is none of the websites business. They can protect their business model on the server side, if they need to. But maybe they just need better business models.
Just SSH. Every public facing piece of software (I.e. a web interface) adds more complexity for misconfiguration or security vulnerabilities.
You can mount you remote filesystem locally and use your local file manager and text editors to manage most tasks. If you use ansible you can make changes to a local configuration and deploy the state to the server without needing to run anything special on the server side. It is especially effective if you also run docker.
And for monitoring I usually just have a tmux with btop running. Which is fine if you don't need long term time series data, then you might want to look at influxdb/grafana - but even those I would run locally behind a firewall, with the server reporting the data to the database.