Sort ref ordering within components

[brynpickering]

#update

Fixes #616

Summary of changes in this pull request

• Rebuild references in the order they appear in the backend model dataset, which is the same order they are added to it (so expressions that depend on previous expressions will be built in the correct order).
• Remove investment cost from demands in the dummy test model - this leads to a change in shadow prices which I have updated, but there's no way of knowing if the new shadow prices (or, indeed, the old ones) are correct.
• Force verbose strings on rebuilt backend model components, if verbose strings have already been triggered.

Reviewer checklist

• Test(s) added to cover contribution
• Documentation updated
• Changelog updated
• Coverage maintained or improved

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 95.34%. Comparing base (872978d) to head (8879a79).
Report is 44 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #623      +/-   ##
==========================================
- Coverage   95.85%   95.34%   -0.52%     
==========================================
  Files          24       24              
  Lines        3619     3650      +31     
  Branches      788      734      -54     
==========================================
+ Hits         3469     3480      +11     
+ Misses         86       82       -4     
- Partials       64       88      +24     

Files with missing lines	Coverage Δ
src/calliope/backend/backend_model.py	98.07% <100.00%> (+0.16%)	⬆️
src/calliope/backend/pyomo_backend_model.py	95.59% <100.00%> (+0.17%)	⬆️

... and 3 files with indirect coverage changes

[irm-codebase]

Few comments on the tests.

[irm-codebase]

A tad verbose... I'd say the stages before update_parameter are not needed (or a separate test?).

Suggestions:

• I'd keep the docstring to a short sentence, like "Check that parameter re-definition propagates to constraints and obj. function"
• re-arrange so that asserts are done once, not inside for loops.
• avoid defining functions inside the test.
• Change 0 (which could be a default) to something rare. I like 0xDEADBEEF, a common debug integer.

[brynpickering]

Thanks for the comments @irm-codebase - updated!

[irm-codebase]

Great changes! Approved.