* Rename old get_scalar_constant_value to get_underlying_scalar_constant

Bumps [pypa/cibuildwheel](https://github.com/pypa/cibuildwheel) from 2.12.0 to 2.12.1.
- [Release notes](https://github.com/pypa/cibuildwheel/releases)
- [Changelog](https://github.com/pypa/cibuildwheel/blob/main/docs/changelog.md)
- [Commits](https://github.com/pypa/cibuildwheel/compare/v2.12.0...v2.12.1)

---
updated-dependencies:
- dependency-name: pypa/cibuildwheel
  dependency-type: direct:production
  update-type: version-update:semver-patch
...
Signed-off-by: dependabot[bot] <support@github.com>

Bumps [pypa/gh-action-pypi-publish](https://github.com/pypa/gh-action-pypi-publish) from 1.5.0 to 1.8.1.
- [Release notes](https://github.com/pypa/gh-action-pypi-publish/releases)
- [Commits](https://github.com/pypa/gh-action-pypi-publish/compare/v1.5.0...v1.8.1)

---
updated-dependencies:
- dependency-name: pypa/gh-action-pypi-publish
  dependency-type: direct:production
  update-type: version-update:semver-minor
...
Signed-off-by: dependabot[bot] <support@github.com>

JAX behavior at the boundary changed and no longer returns np.inf like scipy does. Fixed by evaluating at a non-boundary value.

Without extracting the underlying numeric value of a constant TensorVariable `length`, `stop == length` would always fail.

* Rename sgn to sign

Bumps [actions/download-artifact](https://github.com/actions/download-artifact) from 2 to 3.
- [Release notes](https://github.com/actions/download-artifact/releases)
- [Commits](https://github.com/actions/download-artifact/compare/v2...v3)

---
updated-dependencies:
- dependency-name: actions/download-artifact
  dependency-type: direct:production
  update-type: version-update:semver-major
...
Signed-off-by: dependabot[bot] <support@github.com>

Empty DimShuffles would return a scalar instead of an array, which can then lead to errors in ops that expect an array.

It doesn't make sense to include `fast_run` if `fast_compile` mode is being used. Some rewrites such as the FusionOptimizer are not compatible with `fast_compile` mode which prevents the creation of C thunks. The FusionOptimizer has no way of knowing this is the case, and assumes it is safe to return Composites with more than 32 operands, even though that's not the case with the Python perform method.

This was not an issue in my local machine, but failed on the Github CI. It could be due to compiler optimizations. Case 69 used to look like this:

```python
 Elemwise{Composite{(i0 * tan(i0) * tan(i0) * i1)}} [id C]
 |x [id A]
 |x [id A]
 ```

 And now looks like this
 ```python
 Elemwise{Composite{(i0 * tan(i0) * tan(i0) * i0)}} [id C]
 |x [id A] [None]
 ```