Fix controlled CU QPY serialization

[TSS99]

Summary

Fixes #11686. A circuit containing a controlled CUGate (for example
CUGate(...).control()) crashed on qpy.load with
TypeError: label expects a string or None.

Cause

CUGate carries four parameters (θ, φ, λ, γ), but its base gate is a
three-parameter UGate. When another control was added, the generic control
helper kept the three-parameter UGate as the base while the resulting
controlled gate exposed all four CUGate parameters. On reload QPY rebuilt the
base as UGate(θ, φ, λ, γ), passing the fourth value γ into the label
argument.

Fix

For a controlled CUGate, the base gate now explicitly represents
e^{iγ}·U(θ, φ, λ) (a UGate with a global phase of γ), so it genuinely has
four parameters. QPY round-trips it and the circuit operator is preserved.

Tests

• test_controlled_cu_gate in test_circuit_load_from_qpy.py – QPY round-trip
  of a controlled CUGate.
• test_controlled_cugate_base_gate_includes_global_phase in
  test_controlled_gate.py – the base gate equals e^{iγ}·U(θ, φ, λ).

AI/LLM disclosure

• I didn't use LLM tooling, or only used it privately.
• I used the following tools to understand this PR: Claude and ChatGPT

[qiskit-bot]

Thank you for opening a new pull request.

Before your PR can be merged it will first need to pass continuous integration tests and be reviewed. Sometimes the review process can be slow, so please be patient.

While you're waiting, please feel free to review other open PRs. While only a subset of people are authorized to approve pull requests for merging, everyone is encouraged to review open pull requests. Doing reviews helps reduce the burden on the core team and helps make the project's code better for everyone.

One or more of the following people are relevant to this code:

• @Qiskit/terra-core
• @mtreinish

[jakelishman]

This can't be the cleanest way to go about the stated fix: it's neither a change to the CUGate class, nor to QPY. The CUGate/ControlledGate subclass relationship is somehow different to other gates: either that should be fixed directly (which probably has a myriad backwards compatibility concerns), or QPY needs a direct fix.

Certainly, modifying the generic backing of Gate.control (very indirectly) and not the ostensibly problematic CUGate.control is not the answer here.

[TSS99]

@jakelishman , Thanks for the feedback! I've updated CUGate.control to handle the CU-specific base gate and reverted _add_control.py back to its original generic implementation. Both the controlled-CU base-gate and QPY round-trip regression tests are now passing.