Create BCI pipeline for Model PRs on main

[CodeGat]

Description
One of the stages for open PRs (specifically, on the main branch) on models would be to run cheap BCI tests as the PR is updated. These include Code Unit Tests (where they exist) as well as Semantic and Syntax checks. Find some syntax/semantic checkers that can be used as an action.

For pre-merge checking, the full model build test will be run (in which the version at the tip of the feature branch is 'transplanted' onto the relevant dependency image). This is also run for all dependent models, too. So, for example, if mom5 was updated, access-om2 would also have to be updated as well. (Look up how it would work to transplant multiple pieces into a dependency image (eg. using mom5@=git.feature_branch in the access-om2 environment, which needs to build access-om2@=git.master.

Considerations

• How long would this pre-merge checking take?

[CodeGat]

Unassigning myself to this ticket for the time being as other, higher priority work has come up (namely #5 ) after a standup chat with @aidanheerdegen .
State of the ticket: Chatted to Paul and some other people at ACCESS-NRI about Fortran linting standards and conventions, but hadn't got to a consensus on what standard to use, or even what errors to check for (indeed, finding a decent GHA linter is difficult enough).
One thing that did come up is that we don't want to enforce standards on older models - some of them are almost 10 years old and aren't going to be changed.