BarrierOp and Circuit._add_conditional_barrier

[sjdilkes]

Adds a new method to the Circuit object _add_conditional_barrier.
Previously OpType::Barrier was classified as a MetaOp. However, it seemed a bit at odds with other MetaOp, taking a variable number and mix of UnitID, and having special case handling for being added to a Circuit by a user, while other MetaOp can't be.
Given this, this new additional caveat of being to add them as a conditional gate seemed like a step far enough to make them a new child Op BarrierOp. Happy to undo this work and just have as a special case MetaOp if preferred though.

I have made the method "hidden" currently as it is unusual functionality - but obviously we can give it a normal name if we prefer.

[cqc-alec]

Thanks, code looks good. One suggestion I'd make, before merging this, would be to build a wheel and verify that it solves the problem this was designed to solve. Because it does feel like a strange feature -- "conditional barrier" doesn't really have semantic meaning at the circuit level (condition is evaluated at runtime; barrier is a compile-time annotation.)

[sjdilkes]

Thanks, code looks good. One suggestion I'd make, before merging this, would be to build a wheel and verify that it solves the problem this was designed to solve. Because it does feel like a strange feature -- "conditional barrier" doesn't really have semantic meaning at the circuit level (condition is evaluated at runtime; barrier is a compile-time annotation.)

Good idea. I've messaged Dominic to check that the qasm output is valid for their input and messaged Kentaro to make sure that it suitably constructs his desired circuits.

[sjdilkes]

Thanks, code looks good. One suggestion I'd make, before merging this, would be to build a wheel and verify that it solves the problem this was designed to solve. Because it does feel like a strange feature -- "conditional barrier" doesn't really have semantic meaning at the circuit level (condition is evaluated at runtime; barrier is a compile-time annotation.)

Good idea. I've messaged Dominic to check that the qasm output is valid for their input and messaged Kentaro to make sure that it suitably constructs his desired circuits.

Dominic has confirmed that the output qasm is as expected.