Deprecate Bit.index and Bit.register

[kdk]

Summary

Deprecate references from Bit objects to their containing Register.

Details and comments

on hold for #5519 .

[mtreinish]

Overall LGTM 2 questions inline, the one thing I'm not clear on is whether we're just deprecating Qubit.index or this is also deprecating the registers pieces in the constructor. Looking at the code and the release notes it's only about the attributes/properties, but this updates the docstring for Clbit and Qubit to say deprecated.

[mtreinish]

Are we deprecating the constructor or just the access?

[kdk]

Good catch, this is left over from an earlier iteration. Right now just access, constructor will be done when we deprecate implicit register creation.

[kdk]

Reverted in 17406ac .

[mtreinish]

I don't remember how we handled registerless bits in the layout, but won't this potentially miss a virtual bit if it's not in a register?

[kdk]

Correct, fixed in afdffa3 .

[mtreinish]

Suggested change

	via the :meth:`~qiskit.circuit.Qubit.register` or :meth:`~qiskit.circuit.Qubit.index`
	via the :attr:`~qiskit.circuit.Qubit.register` or :attr:`~qiskit.circuit.Qubit.index`

Only merge this if there are no other changes (to avoid ci churn), if not I can fix this in #6075

[kdk]

Fixed in afdffa3 .

[mtreinish]

LGTM, it'd be good to have some gate_map visualization tests without a register but that can come in a follow up (especially since those tests need to be refactored anyway)