opendp 0.14.2-dev.20260401.2

A library of differential privacy algorithms for the statistical analysis of sensitive private data.
These snippets are built into the reference API documentation and run as tests, although the path is indirect:

## For Python or R:

- From the `rust` directory, runs `cargo build --all-features` to rebuild generated code.
(`cargo build --features bindings` should work but currently fails. See https://github.com/opendp/opendp/issues/1343.)

## For Python:

- `python.rs` generates Python source code, and if a function name matches a file name with an `.rst` extension here, it will be included as part of the docstring. (Note that if there's a typo in the snippet filename the file will be silently ignored.)
- All Python tests are run by a single invocation of pytest. The `.pytest.ini` file configures two ways that doctests are run:
  - Doctests in Python source, including the generated sources which include these snippets, are run by `--doctest-modules`
  - Doctests in `.rst` files under `docs/` (but not `.rst` files under `rust/`!) are run by `--doctest-glob '*.rst'`
- Globals for the doctests are defined by `python/conftest.py`.

When debugging tests, rather than going through a full build of Python sources, I've found it easier to just use global search and replace to make edits to the snippet and the generated python at the same time.

## For R:

- `r.rs` generates R source code, and if a function name matches a file name with an `.R` extension, the snippet will be included in the generated source code.
- For now, the `then_` function docs use the snippet of the corresponding `make_` function.
- `tools/r_stage.sh -d` will rebuild the R docs from the previously generated R source code.
- In contrast to Python doctests, in general there are no assertions on expected output in the snippets themselves: Instead, by default, the R docs build will run examples, and include output in generated documentation, but fail if an example has an error.