Changes required to use CABLE as a library for ESM1.6

[Whyborn]

Associated PR in ESM1.6 that shows bitwise compatibility with non-library build: ACCESS-NRI/ACCESS-ESM1.6#156
UM PR that removes the CABLE code: https://github.com/ACCESS-NRI/UM7/pull/176

The implementation in the UM7 spack package feels quite clunky but I couldn't think of a better way to do it. Happy to hear suggestions on how to improve it.

[Whyborn]

@harshula Could I get your assistance debugging this failure: I'm trying to work out the CABLE CI failure- says cannot satisfy [email protected], but it builds with no problems in this ESM1.6 PR which uses that spec. I'm expecting the UM7 build failure, as the CI will be trying to build with CABLE as library while still having it as source code, UM7 PR is here.

edit: Worked out the problem, I was misusing the with when functionality.

[Whyborn]

@penguian as per the conversation in the ESM1.6 meeting this morning, the GCC build failures should not block merging. Further, we've removed the GCC build from UM7's build CI.

[SeanBryan51]

"Add UM7 intel specific CI to bypass gcc" - I would have thought we want to keep the GCC test case, do you know why the GCC build was failing?

Oops, sorry I realised that was intentional.

[SeanBryan51]

Approving, no need to address my suggestion to expose CABLE specifics through a single module in this PR since its a pretty big change. But it would be great if that was put in later IMO.

[SeanBryan51]

I don't have write access, @harshula could you give this a review?

[harshula]

HI @SeanBryan51 , I've given you write access.

[penguian]

Try changing ESM1.6 to access-esm1.6

[Whyborn]

I've pushed the changes- the CI will fail on CABLE until this CABLE PR is merged (I think?).

[penguian]

Looks OK to me.

[harshula]

Hi @Whyborn & @penguian ,

re: #372

@penguian did extensive testing of with when(). We should only use it in the member variable declaration area of the class for conditional depends_on() calls.