Fixes issues with SLURM 24.11.5

[allenwang28]

Fixes: #1101

Specific changes:

• Handle job_resources=None for pending jobs (fixing the AttributeError crash)
• Support both heterogeneous (+) and regular (.) job ID formats in sacct
• Add SLURM 24.11+ nodes.allocation parsing for running jobs
• Improve state parsing to handle truncated (CANCELLED+) and verbose states
• Add unit tests for new compatibility scenarios

On a broader note, while this fixes immediate SLURM 24.11.5 issues, nested compatibility checks without proper SLURM versioning in torchx seems difficult to maintain.

Test plan:

• Manually on a cluster with 24.11.5
• Added a few unit tests for the failed test conditions:

pytest torchx/schedulers/test/slurm_scheduler_test.py::SlurmSchedulerTest -vs

[facebook-github-bot]

@kiukchung has imported this pull request. If you are a Meta employee, you can view this in D81938651.