Back out "Improve asset management"

[pytorchbot]

Summary:
Original commit changeset: 46d2f763787b

Original Phabricator Diff: D80715432

Since D80715432 landed, we also landed D81256207 to add more logging. To prevent conflicts, we backout in 3 steps:

• In diff 1/3, we backout D81256207, which added more logging
• In diff 2/3, we backout D80715432, which is the problematic diff we're trying to revert
• In diff 3/3, we introduce some of the logging we backed out in diff 1/3

Differential Revision: D82581443

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/14569

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 5 New Failures, 11 Pending

As of commit a297261 with merge base e0dda90 ():

NEW FAILURES - The following jobs have failed:

• pull / test-llama-lora-linux / linux-job (gh)
  RuntimeError: Command docker exec -t db2a2ff6f00ae1a3609ebb7a111cceb87e33757b6fc5740753fa62bd34aa8f6e /exec failed with exit code 1
• pull / test-llama-runner-linux (fp32, xnnpack+custom+qe, linux.arm64.2xlarge, executorch-ubuntu-22.04-gc... / linux-job (gh)
  AttributeError: '_OpNamespace' 'llama' object has no attribute 'sdpa_with_kv_cache'
• pull / test-llama-runner-linux (fp32, xnnpack+custom+quantize_kv, linux.arm64.2xlarge, executorch-ubuntu... / linux-job (gh)
  AttributeError: '_OpNamespace' 'llama' object has no attribute 'sdpa_with_kv_cache'
• pull / test-llama-runner-linux (fp32, xnnpack+quantize_kv, linux.arm64.2xlarge, executorch-ubuntu-22.04-... / linux-job (gh)
  AttributeError: '_OpNamespace' 'llama' object has no attribute 'sdpa_with_kv_cache'
• pull / unittest-nxp-neutron / linux-job (gh)
  test_split_group_convolution__2d_1_group_3

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[GregoryComer]

Manually rebasing as there are some minor merge conflicts in ETCoreMLModelManager.mm. Hopefully low risk as conflicts were mainly comments. I double-checked and this specific pick PR is ordered properly with the reverts on main branch. If CI passes, I'll merge.