Address review feedback: add log_cdf equation and use \frac

- Add equation for Normal.log_cdf with note about jax.scipy.stats.norm.logcdf implementation
- Replace \tfrac with \frac in Cauchy.icdf equation

Addresses review comments from @Qazalbash in PR #2188

Author: kimjune01

numpyro/distributions/continuous.py

@@ -439,7 +439,7 @@ def icdf(self, q: ArrayLike) -> ArrayLike:
 
         .. math::
            F^{-1}(q; x_0, \gamma) = x_0 + \gamma \tan\!\left[\pi\!\left(q
-           - \tfrac{1}{2}\right)\right]
+           - \frac{1}{2}\right)\right]
 
         :param q: Probability value in :math:`[0,1]`.
         :type q: ArrayLike
@@ -2736,6 +2736,12 @@ def cdf(self, value: ArrayLike) -> ArrayLike:
     def log_cdf(self, value: ArrayLike) -> ArrayLike:
         r"""Log of the cumulative distribution function.
 
+        .. math::
+           \log F(x; \mu, \sigma) = \log\!\left(\frac{1}{2}\left[1 + \mathrm{erf}\!\left(
+           \frac{x - \mu}{\sigma\sqrt{2}}\right)\right]\right)
+
+        Implementation uses :func:`jax.scipy.stats.norm.logcdf`.
+
         :param value: Value to evaluate.
         :type value: ArrayLike
         """