Amend PCPhase operator to support RZ convention

[Jaybsoni]

Context:
PCPhase gates were designed to be primarily used in the QSVT workflow. Swapping to this convention will allow them to be more directly applied to the workflow without having to keep track of additional global phases.

Description of the Change:
In the previous implementation of PCPhase(phi, dim, wires), the first dim number of entries along the diagonal in the matrix were phase shifted by e^(i * phi) while the remaining 2**n - dim entries of the diagonal were left alone (set to 1).

This convention is swapped to one in which the remaining 2**n - dim entries of the diagonal were set to e^(-i * phi). Note, these two conventions only differ by a global phase.

Benefits:
By absorbing the global phase into the representation, we get a cleaner implementation of QSVT which doesn't have to track the global phase as strictly when converting between representations. This convention is also identical when using trained QSP angles which use RZ rotations as their projector controlled phase gates.

[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 #3926 (867a926) into add_qsvt_functionality (f41261c) will increase coverage by 0.02%.
The diff coverage is 100.00%.

@@                    Coverage Diff                     @@
##           add_qsvt_functionality    #3926      +/-   ##
==========================================================
+ Coverage                   99.71%   99.73%   +0.02%     
==========================================================
  Files                         340      345       +5     
  Lines                       29681    30320     +639     
==========================================================
+ Hits                        29596    30241     +645     
+ Misses                         85       79       -6     

Impacted Files	Coverage Δ
pennylane/ops/qubit/parametric_ops_multi_qubit.py	100.00% <100.00%> (ø)

... and 73 files with indirect coverage changes

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

[obliviateandsurrender]

Thanks @Jaybsoni! Just leaving a few comments. Another minor thing I would like to clarify regarding "these two conventions only differ by a global phase" would be that isn't the new phi in your implementation twice the phi in your old one? Not sure if that causes any problems w.r.t its domain.

[Jaybsoni]

[sc-34913]

[obliviateandsurrender]

Looks good. Is the dipole moment CI failure related to the recent change in pyscf? We had pushed a fix in the master branch for this.

[Jaybsoni]

Looks good. Is the dipole moment CI failure related to the recent change in pyscf? We had pushed a fix in the master branch for this.

I am not sure, let me try merging in master to see if that fixes it?

[soranjh]

Looks good. Is the dipole moment CI failure related to the recent change in pyscf? We had pushed a fix in the master branch for this.

I am not sure, let me try merging in master to see if that fixes it?

Yes, this issue has been fixed here.