Add decomposition to Controlled

[albi3ro]

Currently, ControlledOperation created by ctrl has some very basic decomposition behaviour. But it at least has more decomposition behaviour than the non-integrated Controlled.

This PR makes Controlled reach feature-parity with ControlledOperation and will allow us to replace it.

[github-actions]

Hello. You may have forgotten to update the changelog!
Please edit doc/releases/changelog-dev.md with:

• A one-to-two sentence description of the change. You may include a small working example for new features.
• A link back to this PR.
• Your name (or GitHub username) in the contributors section.

[codecov]

Codecov Report

Merging #2938 (2bce582) into master (8c15817) will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master    #2938   +/-   ##
=======================================
  Coverage   99.65%   99.65%           
=======================================
  Files         266      266           
  Lines       22332    22364   +32     
=======================================
+ Hits        22255    22287   +32     
  Misses         77       77           

Impacted Files	Coverage Δ
pennylane/ops/op_math/controlled_class.py	100.00% <100.00%> (ø)
pennylane/ops/qubit/matrix_ops.py	100.00% <100.00%> (ø)
pennylane/ops/qubit/non_parametric_ops.py	100.00% <100.00%> (ø)
pennylane/ops/qubit/parametric_ops.py	100.00% <100.00%> (ø)

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[AlbertMitjans]

I don't have much time for a full review, but noticed the following:

>>> qml.ops.Controlled(qml.PauliRot(3, "XX", [0, 4]), 2).decomposition()
[C(Hadamard)(wires=[2, 0]),
 C(Hadamard)(wires=[2, 4]),
 C(MultiRZ)(3, wires=[2, 0, 4]),
 C(Hadamard)(wires=[2, 0]),
 C(Hadamard)(wires=[2, 4])]

Shouldn't C(MultiRZ) be decomposed as well?

[dwierichs]

Looks great, @albi3ro! 🎉 Just had a few minor questions and a tiny suggestion.
The new _controlled syntax using a return value looks much better!

[albi3ro]

@AlbertMitjans I want to add more decomposition algorithms in the future, but that would take a bit more work than just translating over what ControlledOperation already does.

[dwierichs]

Looks great @albi3ro 🎉
Just the disordered wires test case question is open from my side.

[Jaybsoni]

Looks good @albi3ro, I can see now how annoying these in place inversions are haha!