Fuel gauge cmake library structure incorrect

[matt-wood-ct]

Describe the bug

Fuel gauge drivers are creating a library per driver rather then one library for the fuel_gauge group with all sources included. This is not inline with the rest of the drivers in zephyr, this was a minor detail missed during the reviews on the original introduction.

The issue with this structure is that its is not properly extensible out of tree, normal out of tree drivers utilize zephyr_library_amend(), this cannot be used on fuel_gauges as there is no library called that to amend.

Edit:
Drafted pull request: #93836

Regression

• This is a regression.

Steps to reproduce

N/A structural issue, reproduction not relevant

Relevant log output

N/A

Impact

Functional Limitation – Some features not working as expected, but system usable.

Environment

N/A

Additional Context

Discussed in Discord in advance of raising issue, @bjarki-andreasen confirms this is a mistake likely missed during initial reviews.

[teburd]

You should be able to create a new library for your driver much like the in tree drivers.

Sensor drivers work the same way, each one is built as a distinct zephyr_library() so I'm not sure what problem exactly you are running across.

zephyr_library_amend() shouldn't be needed.

[matt-wood-ct]

You should be able to create a new library for your driver much like the in tree drivers.

Sensor drivers work the same way, each one is built as a distinct zephyr_library() so I'm not sure what problem exactly you are running across.

zephyr_library_amend() shouldn't be needed.

What is correct then? this behaviour in fuel_gauge and sensor or whats done in the other driver types?

Why is there an inconsistency between different driver types?

Is this just a discussion of scale? does sensor deserve to have libraries per sensor because there are so many?
or perhaps is it because sensor has some "core" files like the sensor_shell and sensor_handlers which should be packaged separately from the actual sensor drivers?

I assume the fact that sensor had a different structure is probably the reason why fuel_gauge doesn't have the structure correct as it was spun out of the sensor drivers originally.

This feels too petty to push up to tsc but in my opinion the cmake structure should really be well defined and consistent.

I'm happy to do the PR, but if this is a hard no I'll just save the effort.

[teburd]

only the build system maintainer would be able to really say one way or another on this one, @tejlmand or maybe @nordicjm would likely be the best folks to discuss with. I'm not so sure this could be considered a bug given some things in zephyr do it one way or another. Neither approach seems wrong or right to me offhand.