config recommender tool for CLI

[themattinthehatt]

Adds litpose recommend <dataset_path>, which analyzes a lightning-pose dataset (labeled
CSV(s), optionally unlabeled videos in <dataset_path>/videos) and available GPU hardware,
prints a rationale-annotated recommendation report, and always writes a complete, ready-to-use
config YAML.

• Recommends model_type (single-view heatmap vs. heatmap_multiview_transformer for
  multi-view datasets, auto-detected from multiple CSVs in the directory), backbone
  (vits_dino), per-side image resize dims, train_batch_size, max_epochs, augmentation
  pipeline (interactively prompts whether the dataset is a top-down freely-moving setup),
  optimizer, and losses_to_use (temporal/pca_singleview when unlabeled videos are present
  and there are enough labeled frames to fit a PCA subspace).
• train_batch_size is GPU-VRAM-aware
• Batch size is further adjusted for context models and multiview models, and for
  semi-supervised training.

[codecov]

Codecov Report

❌ Patch coverage is 94.82759% with 18 lines in your changes missing coverage. Please review.
✅ Project coverage is 92.94%. Comparing base (97f536f) to head (33cf2ef).

Files with missing lines	Patch %	Lines
lightning_pose/utils/recommender.py	95.23%	14 Missing ⚠️
lightning_pose/cli/commands/recommend.py	92.59%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #489      +/-   ##
==========================================
+ Coverage   92.82%   92.94%   +0.11%     
==========================================
  Files          60       62       +2     
  Lines        5620     5968     +348     
==========================================
+ Hits         5217     5547     +330     
- Misses        403      421      +18     

Files with missing lines	Coverage Δ
lightning_pose/cli/commands/__init__.py	100.00% <ø> (ø)
lightning_pose/cli/commands/recommend.py	92.59% <92.59%> (ø)
lightning_pose/utils/recommender.py	95.23% <95.23%> (ø)

🚀 New features to boost your workflow:

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