EstimatorV2 updates

[kt474]

Summary

There are a lot of files copied over from qiskit, they should be replaced after the next qiskit release.

Simple job example:

Details and comments

Fixes #1314

[coveralls]

Pull Request Test Coverage Report for Build 7835592117

• -9 of 59 (84.75%) changed or added relevant lines in 15 files are covered.
• 5 unchanged lines in 2 files lost coverage.
• Overall coverage increased (+0.5%) to 79.982%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
qiskit_ibm_runtime/utils/estimator_result_decoder.py	1	3	33.33%
qiskit_ibm_runtime/utils/json.py	15	18	83.33%
qiskit_ibm_runtime/sampler.py	5	9	55.56%

Files with Coverage Reduction	New Missed Lines	%
qiskit_ibm_runtime/utils/estimator_result_decoder.py	2	50.0%
qiskit_ibm_runtime/utils/json.py	3	72.2%

Totals
Change from base Build 7833396730:	0.5%
Covered Lines:	5262
Relevant Lines:	6579

---

💛 - Coveralls

[ihincks]

BaseEstimatorV2 (and all supportive container classes) has now been merged into qiskit's main branch

[kt474]

@jyu00 @gadial
Should be ready for initial review, I left a couple questions in about things I wasn't sure about

[gadial]

Generally LGTM. I believe we should have more tests, and my main concern is the large amount of copy-pasting of container data from qiskit, even if we believe it's not subject to change. However, if we have no way around it I have no objections to merging this PR.

[gadial]

I suggest we treat this similarly to other complex data types, e.g. pass a the following dict:

        if isinstance(obj, EstimatorPub):
            return {
                "__type__": "EstimatorPub",
                "__value__": {
                    "circuit": obj.circuit,
                    "observables": obj.observables,
                    "parameter_values": obj.parameter_values,
                    "precision": obj.precision,
                },
            }

However, I think we should delay that until we start using the EstimatorPub from qiskit itself.

[gadial]

The test suite is currently relatively small; perhaps we can use the test suite in qiskit:

https://github.com/Qiskit/qiskit/blob/main/test/python/primitives/test_estimator.py

See the SamplerV2 PR where I do something similar for Sampler:

https://github.com/gadial/qiskit-ibm-runtime/blob/8102831d23fba61fbe294770accbe093f32929a6/test/integration/test_sampler_v2.py

[jyu00]

Now that BaseEstimatorV2 and container classes are all merged into Qiskit main, I think it'd make more sense to just include qiskit main (then rc1 when it's released) in this branch

[kt474]

Now that BaseEstimatorV2 and container classes are all merged into Qiskit main, I think it'd make more sense to just include qiskit main (then rc1 when it's released) in this branch

Removed all of the copied files and using the main branch of qiskit now. I believe the build is failing because the CI is not using the main branch and the qiskit version is being overwritten by another dependency.

[gadial]

Now that BaseEstimatorV2 and container classes are all merged into Qiskit main, I think it'd make more sense to just include qiskit main (then rc1 when it's released) in this branch

Removed all of the copied files and using the main branch of qiskit now. I believe the build is failing because the CI is not using the main branch and the qiskit version is being overwritten by another dependency.

This looks like mthree dependency, but I don't see why mthree is included in requirements-dev.txt at all.

[kt474]

Lint is now failing because qiskit.test was removed from qiskit 1.0 (Have a separate PR to take care of that #1292, too many moving parts 🫠)

Ran unit tests locally - tests are passing except test_parameters_vals_kwvals (which I think is from a bug in BindingsArray.coerce() - fixed Qiskit/qiskit#11645

[jyu00]

This can be a separate PR, but v2 primitives run() method now returns an instance of BasePrimitiveJob, which does not inherit from JobV1. RuntimeJob needs to be updated to inherit this (plus JobV1 to support backend.run still, if the same class is used).

[jyu00]

@kt474 Now that the 2 blockers have been merged, can you update this PR so we can get it merged? We'll need more tests, but that can go into a separate PR.