Release `etn` API update

[PietrH]

I'll use this issue to track outstanding tasks for the next big etn release.

Bart-v: FYI: Paul is off until 2025-09-29 & Stijn is off until 2025-10-14
We probably shouldn't release until #395 has been discussed, and when VLIZ IT is ready to fix any issues that might come up with the release.

• Create pull request to main
• Clear any merge conflict to main
• Discuss next version number
• Ensure RStudio Server 4 is open to users
• Ensure etnservice production release is most recent (0.5.0)
• Consider bumping etnservice version number to v1
• Ensure I have access to RStudio Server 4
  I've had trouble logging since 2025-10-01
• Talk about API instability with VLIZ
• Discuss about message in ETN newsletter
• build the pkgdown website

Discussion points

Version number

The changes in this release are major, and breaking.

• The connection argument still exists, but is unused and returns a warning.
• The way to pass credentials to the package has changed (you enter them when prompted, on every call, or store them on a profile, session or system level)
• There is new functionality, new arguments, and functions return different results than they used to (due to changes in queries and views), especially for detections
• Much larger detection queries are now possible
• There is a dependency on R4
• There is a dependency on arrow

Because of this, I feel like a major version bump to etn v3.0.0 might be warranted.

Alternatively, the next version number would be etn v2.3.0

The development branch is named v2.3.1. This just kind of happened really. And is of no consequence to this decision.

The connection argument

Currently the connection argument is still the first argument in almost all functions, if you pass a value to it, you will get a warning.

Should we just formally remove it? So you can write: get_acoustic_projects("2004_Gudena") instead of get_acoustic_projects(acoustic_project_code = "2004_Gudena")?

I could add ellipsis ... to the functions where I remove the argument, which would allow me to still warn users if they pass it as a named argument: connection = my_con, but this wouldn't stop old scripts form passing connection arguments unnamed: get_acoustic_projects(my_con, "2004_Gudena").