Detect duplicates in QuantumCircuit.compose

[jakelishman]

Summary

Previously, QuantumCircuit.compose would silently attempt to compose multiple qubits onto the same state, which would fail in weird ways if any multi-qubit instructions were acting on these collapsed qubits.

Details and comments

Fix #4970.

[qiskit-bot]

One or more of the the following people are requested to review this:

• @Qiskit/terra-core

[coveralls]

Pull Request Test Coverage Report for Build 7448349417

Warning: This coverage report may be inaccurate.

We've detected an issue with your CI configuration that might affect the accuracy of this pull request's coverage report.
To ensure accuracy in future PRs, please see these guidelines.
A quick fix for this PR: rebase it; your next report should be accurate.

• 0 of 0 changed or added relevant lines in 0 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage decreased (-0.004%) to 87.574%

---

Totals
Change from base Build 7416093664:	-0.004%
Covered Lines:	59484
Relevant Lines:	67924

---

💛 - Coveralls

[Cryoris]

LGTM, now the raised error makes a lot more sense 👍🏻

[jakelishman]

In retrospect, I'll remove this from "stable backport", because the bug's been open for ages without huge complaints, and we're close to 1.0 anyway, so there's little need to increase the risk of the 0.45.2 release.

[jakelishman]

The new test here caught a sort-of bug in #11214 where there was an early return for empty data in the composed operation without checking the qubits and clbits (the errors weren't raised). Alone that's not much of a problem, but the bigger issue that it turned up was that global-phase and calibrations were added to the dest circuit before error checking, which meant that if the composition raised an exception, we could be leaving the dest in an partially mutated state, which could cause issues for circuit-reuse after exception handling if we were doing an in-place composition.

[Cryoris]

Nice catch 👍🏻 there seem to be some other modifications to dest happening before, such as changing the dt unit and the duration. Should we make sure these are also modified only after the check?

[jakelishman]

The duration and unit question came up in #11214 and the result was a resounding "uh" - we don't really know what those are for at a circuit level, nor what meaning they're supposed to have (what if a circuit is made of incompatible durations with incompatible units?).

Seems easy enough to change, at any rate - done in 48116a. While it appears like I moved it after some possible return paths, those return paths all (necessarily) call QuantumCircuit.append, which will zero out the duration and unit as well.

[Cryoris]

LGTM thanks for the updates!