Add P.UNCONSTRAINED condition in remove size one mesh axis

[NuojCheng]

Description

For MaxText.sharding.remove_size_one_mesh_axis function we always assume the given partition spec exists in mesh, but using jax.sharding.PartitionSpec.UNCONSTRAINED violates this setting. This PR fixes this connor case to avoid possible bugs (mostly using pipeline parallelism + debug_sharding=True).

Tests

Before change: https://paste.googleplex.com/6298233649364992

After change: https://paste.googleplex.com/6317784340496384

Checklist

Before submitting this PR, please make sure (put X in square brackets):

• I have performed a self-review of my code. For an optional AI review, add the gemini-review label.
• I have necessary comments in my code, particularly in hard-to-understand areas.
• I have run end-to-end tests tests and provided workload links above if applicable.
• I have made or will make corresponding changes to the doc if needed, including adding new documentation pages to the relevant Table of Contents (toctree directive) as explained in our documentation.

[codecov]

Codecov Report

❌ Patch coverage is 0% with 1 line in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/MaxText/sharding.py	0.00%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!

[richjames0]

lgtm