Qiskit / qiskit-ibm-runtime / 29942340657

88%

main: 88%

push

github

Use BaseModel for options_models (#3042)

### Summary

Proof of concept for using `BaseModel` for new program options at
`options_models`:
* Add `BaseOptionsModel` as the pure pydantic base class for options
   * replicates `update()` and `asdict()` for backwards compatibility
   * trims the `dir()` return for auto-completion conciseness
* Revise existing `option_models`
* inherit from `BaseOptionsModel` instead of using `pydantic.dataclass`
   * avoid `Field(default_factory=dict)` pattern, for readability
* use `Annotated` pattern for validators (replacing validators for
non-serializable classes with `InstanceOf`)
* revise docstrings (length, rely on types for users discovering
options, other misc tweaks)
   * focus on keeping them as containers and lean
      * conversion responsibilities moved to `converters.py`
      * revise class-level constants  
* Adjust the rest of the codebase
   * replace `asdict()` with `.model_dump()`
   * other misc adjustments
* Tests
   * add tests for the base model
   * adjust existing tests

### Details and comments

Fixes #3041

### AI/LLM disclosure

- [x] I didn't use LLM tooling, or only used it privately.

170 of 171 new or added lines in 29 files covered. (99.42%)

8985 of 10256 relevant lines covered (87.61%)

6.95 hits per line