Robust Gaussian Processes via Relevance Pursuit (pytorch#2608)

Summary:
Pull Request resolved: pytorch#2608

This commit adds the implementation of the [Robust Gaussian Processes via Relevance Pursuit](https://arxiv.org/pdf/2410.24222) models and algorithms of the NeurIPS 2024 article.

Differential Revision: D65343571

botorch/models/gpytorch.py

@@ -387,19 +387,19 @@ def _apply_noise(
                 obs_noise = observation_noise.squeeze(-1)
             mvn = self.likelihood(
                 mvn,
-                X,
+                [X],
                 noise=obs_noise.expand(noise_shape),
             )
         elif isinstance(self.likelihood, FixedNoiseGaussianLikelihood):
             # Use the mean of the previous noise values (TODO: be smarter here).
             observation_noise = self.likelihood.noise.mean(dim=-1, keepdim=True)
             mvn = self.likelihood(
                 mvn,
-                X,
+                [X],
                 noise=observation_noise.expand(noise_shape),
             )
         else:
-            mvn = self.likelihood(mvn, X)
+            mvn = self.likelihood(mvn, [X])
         return mvn
 
     def posterior(