Fix bug of base_height_l2 reward function

[xliu0105]

Description

When using the manager-based env, configuring the terrain to be rough, using 'base_height_l2' in the reward term, and using raycaster to adjust the target height to account for the terrain, the program will calculate 'Nan' or 'inf', or suddenly calculate a singular value at a certain moment, causing the code to crash. Therefore, the following modifications are made:

• Determine whether there are Nan, inf or singular values
• Calculate the number of ray_hits for Nan, inf and singular values ​​in each environment respectively
• Remove Nan, inf and singular values and calculate the average of the valid values
• If all ray_hits values ​​in an environment are Nan, inf or singular values, use its root height as target_hight

NOTE: I think the cause of this bug may be in the underlying structure of Isaaclab or Isaacsim. The modifications I made can only avoid training failures due to data anomalies. I hope the maintainers can investigate the underlying cause of this problem.

Fixes #1928

Type of change

• Bug fix (non-breaking change which fixes an issue)

Screenshots

None

Checklist

• I have run the pre-commit checks with ./isaaclab.sh --format
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• I have updated the changelog and the corresponding version in the extension's config/extension.toml file
• I have added my name to the CONTRIBUTORS.md or my name already exists there