Auto-select GPUs

[gitttt-1234]

This PR adds functionality to automatically select GPUs with the most available memory when the number of training devices specified by the user is less than the total GPUs available. In this case, we query the memory utilization of each GPU, sort them by free memory, and allocate the devices with the highest availability.

[codecov]

Codecov Report

❌ Patch coverage is 86.66667% with 4 lines in your changes missing coverage. Please review.
✅ Project coverage is 93.49%. Comparing base (ff91433) to head (5a87d7c).
⚠️ Report is 39 commits behind head on main.

Files with missing lines	Patch %	Lines
sleap_nn/training/utils.py	83.33%	4 Missing ⚠️

❌ Your patch status has failed because the patch coverage (86.66%) is below the target coverage (95.00%). You can increase the patch coverage or adjust the target coverage.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #333      +/-   ##
==========================================
- Coverage   95.28%   93.49%   -1.80%     
==========================================
  Files          49       49              
  Lines        6765     6962     +197     
==========================================
+ Hits         6446     6509      +63     
- Misses        319      453     +134     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.