[AUD-002][Critical] Fix MRAC state-error indexing for non-square dimensions

[gabrielsantosphilips]

Audit finding

• ID: AUD-002
• Status: Verified defect
• Severity: Critical
• Confidence: High
• Audited revision: 2f479320d805a1f9f35ebe4afaaeeded48913a94

Problem

ModelReferenceAdaptiveControl::ComputeControl() loops over InputSize but indexes e.at(i, 0) where e has StateSize rows. A valid ModelReferenceAdaptiveControl<float, 1, 2> aborts on the second input; without bounds assertions this is undefined memory access. When InputSize < StateSize, state-error components are silently ignored.

Source:

numerical-toolbox-cpp/numerical/nonlinear_control/ModelReferenceAdaptiveControl.hpp

Lines 81 to 98 in 2f47932

	const StateVector& x, const InputVector& r, T dt)
	{
	xm = xm + (reference.A * xm + reference.B * r) * dt;
	const StateVector e{ x - xm };
	const InputVector u{ thetaX * x + thetaR * r };
	const T scale{ gamma * signB * dt };
	for (std::size_t i = 0; i < InputSize; ++i)
	{
	for (std::size_t j = 0; j < StateSize; ++j)
	thetaX.at(i, j) -= scale * e.at(i, 0) * x.at(j, 0);
	for (std::size_t j = 0; j < InputSize; ++j)
	thetaR.at(i, j) -= scale * e.at(i, 0) * r.at(j, 0);
	}

The current update is also not a dimensionally general MIMO Lyapunov law; it needs an adaptation direction such as B^T P e, rather than the first input-count error entries. AdaptationLaw is stored but not used.

Acceptance criteria

• No state vector is indexed using InputSize unless dimensions guarantee safety.
• The supported MIMO adaptation law is mathematically specified and dimensionally valid.
• Unsupported dimension/law combinations fail at compile time or construction.
• Add float-only TEST_F cases for InputSize > StateSize and InputSize < StateSize.