Refactor MC attitude controller: use module base and Param classes

[bkueng]

• use ModuleBase class and add module documentation
• move class declaration into a separate header
• switch all params from the C API to the C++ Param class
• add some missing orb_unsubscribe's

@dagar this will conflict a bit with #9103. I can do the rebase, should it be accepted and this be merged first.

[dagar]

Don't worry about the conflict, I'm fine either way and this looks more important.

One smaller thing it might be time to discuss is naming. We have a mix of headers with .h (c & c++) and .hpp extensions. Some headers are named to match the module, some named to match the only class they contain.

Personally I have a minor preference for c++ headers that are named to match the containing class, but really the only thing I can't stand is inconsistency. I don't want to get carried about with a detailed style guide, let's just try and get the main contributors roughly on the same page.

[bkueng]

Personally I have a minor preference for c++ headers that are named to match the containing class, but really the only thing I can't stand is inconsistency. I don't want to get carried about with a detailed style guide, let's just try and get the main contributors roughly on the same page.

Same for me. Naming after the class name is not that important, as you can have multiple classes. But the .hpp vs .h suffix is currently a mess and I always wonder which one I should be using. The same goes for camel case vs underscore in filenames. I tend towards .hpp and underscore, so it's easy to distinguish between C/C++ and I think we mostly already use underscores.

[MaEtUgR]

Looks very good on the first glance! I volunteer to switch completely to matrix library as a follow up to this pr. This was the plan for some time already.

[dagar]

I volunteer to switch completely to matrix library as a follow up to this pr. This was the plan for some time already.

@MaEtUgR that would be great! Converting the multicopter controllers should be straightforward (and safe). After that and a few outstanding PRs (#9121, #9104, #9103) we're actually very close to being able to drop mathlib matrix entirely. That will help free up the flash needed for flight tasks on fmu-v2.

[dagar]

@MaEtUgR we don't have to resolve it in this PR, but did you notice this line? We shouldn't carry around commented out code, so it would be good to address at some point.

[MaEtUgR]

@dagar I just saw that it was ignored by @bkueng and you before and assumed everyone knows why it's commented out. After all you commented it: d6e9287 😄

Now I found the error: #9166