Add `LogitNormalVariable` as pre-specified RV

[cdc-mitzimorris]

A logit normal prior is a useful distribution. Currently, it is specified via low-level classes DistributionalVariable and TransformedVariable, e.g.:

iedr_rv = TransformedVariable(
    name="iedr",
    base_rv=DistributionalVariable(
        name="logit_iedr",
        distribution=dist.Normal(
            transformation.SigmoidTransform().inv(0.004),
            0.3,
        ),
    ),
    transforms=transformation.SigmoidTransform(),
)

Instead, it should be possible to specify:

iedr_rv = LogitNormalVariable(name="iedr",  base_name="logit_iedr",  median=0.004,  scale=0.3)

Add class LogitNormalVariable to pyrenew.randomvaraible to make this a first-class PyRenew component:

from pyrenew.randomvariable import DistributionalVariable, TransformedVariable
  import pyrenew.transformation as transformation
  import numpyro.distributions as dist

  def LogitNormalVariable(
      name: str,
      median: float,
      scale: float,
      base_name: str | None = None,
  ) -> TransformedVariable:
      return TransformedVariable(
          name=name,
          base_rv=DistributionalVariable(
              name=base_name or f"logit_{name}",
              distribution=dist.Normal(
                  transformation.SigmoidTransform().inv(median),
                  scale,
              ),
          ),
          transforms=transformation.SigmoidTransform(),
      )

Tests should verify:

• returned object is a TransformedVariable,
• base RV uses dist.Normal(logit(median), scale),
• samples lie in (0, 1),
• names are recorded as expected.