Add RemoveFinalReset pass

[garrison]

Summary

This adds a RemoveFinalReset pass, which removes a Reset instruction if it comes at the end of a qubit wire. It also enables this pass at optimization level 1 and higher.

Details and comments

The new wire cutting code in the Circuit Knitting Toolbox results in circuits where there are Resets in the circuit. In the special case that no qubits are re-used later in the circuit, then all of these Resets are either at the beginning of the circuit or the end of the circuit. There is an existing pass, RemoveResetInZeroState, that removes resets at the beginning of the circuit. This pass is disabled by default in the pass managers (see #10591), as it places assumptions on the state at the beginning of the circuit. This PR creates the other transpiler pass that we need, namely one that removes resets at the end of the circuit. If nothing happens to a qubit wire after, then this removal should have no effect on the circuit. Of course, this does change the result if you then concatenate it with another circuit; I'm assuming this assumption is okay, but if not, it should be removed from the default passes. EDIT: I have removed this from the default passes in 4adc048.

Related CKT issue: Qiskit/qiskit-addon-cutting#452

[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 8347062095

Warning: This coverage report may be inaccurate.

This pull request's base commit is no longer the HEAD commit of its target branch. This means it includes changes from outside the original pull request, including, potentially, unrelated coverage changes.

• For more information on this, see Tracking coverage changes with pull request builds.
• To avoid this issue with future PRs, see these Recommended CI Configurations.
• For a quick fix, rebase this PR at GitHub. Your next report should be accurate.

Details

• 15 of 15 (100.0%) changed or added relevant lines in 3 files are covered.
• 13 unchanged lines in 4 files lost coverage.
• Overall coverage decreased (-0.3%) to 89.316%

---

Files with Coverage Reduction	New Missed Lines	%
qiskit/primitives/containers/estimator_pub.py	1	97.59%
qiskit/primitives/containers/sampler_pub.py	2	97.26%
crates/qasm2/src/lex.rs	4	92.44%
crates/qasm2/src/parse.rs	6	97.62%

Totals
Change from base Build 8320404144:	-0.3%
Covered Lines:	59822
Relevant Lines:	66978

---

💛 - Coveralls

[mtreinish]

It'll probably be a bit more efficient to use the output_map in the dag and look at it's predecessors. This is because dag.op_nodes() iterates over all the instructions in the dag and in a typical circuit there will be far more instructions than qubits and clbits.

Suggested change

	resets = dag.op_nodes(Reset)
	for reset in resets:
	successor = next(dag.successors(reset))
	if isinstance(successor, DAGOutNode):
	dag.remove_op_node(reset)
	for output_node in dag.output_map.values():
	if isinstance(output_node.wire, Qubit):
	pred = next(dag.predecessors(output_node))
	if isinstance(pred, DAGOpNode) and isinstance(pred.op, Reset):
	dag.remove_op_node(pred)

[garrison]

Fixed in f358d46. I wonder if this could potentially be improved in the RemoveResetInZeroState pass too (I modeled my original code on this pass).

[mtreinish]

Yeah, we can probably make a similar change to RemoveResetInZeroState. It looks like it hasn't had any meaningful updates since it was first introduced in #2139.

[garrison]

@mtreinish: I fixed a few things; this is ready to merge now.

[mtreinish]

FWIW, you could make pylint happy if you added a new line between the qiskit import section and this import.