Collapse enzyme wb, ws, h

[lukelowry]

Description

Refactors PhasorDynamics components to use a consistent internal/external variable and residual interface (thank you for the inspo @alexander-novo). This removes the separate wb, ws, and h paths, simplifies Enzyme Jacobian generation, and makes residual ownership clearer without changing model behavior.

Proposed changes

• Add shared external variable and residual storage to Component.
• Split residual evaluation into internal equations and external contributions.
• Consolidate the specialized Enzyme derivative wrappers into generic internal/external forms.
• Update the affected PhasorDynamics models and Jacobian assembly.
• Ignore invalid row and column mappings during sparse Jacobian storage.

Checklist

• All tests pass.
• Code compiles cleanly with flags -Wall -Wpedantic -Wconversion -Wextra.
• The new code follows GridKit™ style guidelines.
• There are unit tests for the new code.
• The new code is documented.
• The feature branch is rebased with respect to the target branch.
• I have updated CHANGELOG.md to reflect the changes in this PR. If this is a minor PR that is part of a larger fix already included in the file, state so.

Further comments

This demonstrates that we don't need to give special treatment to Bus, and we can have Bus inherit from Component, which dramatically simplifies SystemModel and signal orchestration. Personally, this is much easier to document and implement. MOST IMPORTANTLY: this internal/external formulation can allow us to make the residual evaluation order independent, which (I think) is better for partitioning and parallelization.

cc @nkoukpaizan

Draft just so others can see, I may need to break this up

[lukelowry]

@Steven-Roberts this representation might make partitioning easier aswell

[nkoukpaizan]

Thanks @lukelowry for looking at this. Updating these local vector is on our todo list, and this should work, now that we are storing the current injections as local variables. I would still prioritize merging the models you need for the 10K over this change.

[lukelowry]

Thanks @lukelowry for looking at this. Updating these local vector is on our todo list, and this should work, now that we are storing the current injections as local variables. I would still prioritize merging the models you need for the 10K over this change.

I agree, definitely

[lukelowry]

@pelesh I believe this formulation of the system gives me the most flexibility in EMT models. On my end, I would be okay with splitting off from Phasor Dynamics after this.