fix: preserve positional __init__ args through from_config round trips

[gauravch-code]

What does this PR do?

The @register_to_config decorator computed _use_default_values by diffing new_kwargs against init_kwargs, but init_kwargs only captured keyword arguments. Positional args mapped into new_kwargs via zip were never tracked as explicitly provided, so from_config silently reverted them to class defaults.

s = DDIMScheduler(500)
s.config.num_train_timesteps          # 500
DDIMScheduler.from_config(s.config).config.num_train_timesteps  # 1000 (wrong)

The fix tracks positional arg names separately and includes them in the set of explicitly provided parameters when computing _use_default_values.

Fixes #14460

Before submitting

• Did you use an AI agent (Claude Code, Codex, Cursor, etc.) to help with this PR? If so:
  • Did you read the Coding with AI agents guide?
  • Did you share the final self-review notes in the PR description or a comment?
• Did you read the contributor guideline?
• Was this discussed/approved via a GitHub issue? register_to_config mislabels positional __init__ args as _use_default_values, so from_config round trips silently revert them to defaults #14460
• Did you write any new necessary tests?

Who can review?

@sayakpaul @yiyixuxu (general functionalities)

[gauravch-code]

Hi @sayakpaul, could you take a look at this small configuration round-trip fix or point me to the appropriate reviewer? It fixes #14460 by distinguishing positionally supplied arguments from defaulted values and includes a focused regression test; all 12 relevant config tests pass locally. Happy to revise the implementation based on your guidance.