Unified internal implementation for decompositions

[astralcai]

Context:
Right now, preprocess and transforms have two different _operator_decomposition_gen to do recursive long decision trees. This will eventually cause much pain and difficulty, especially when we are ready to have our new graph-based decomposition to do the device preprocess as well.
Since these two methods are doing more or less the same functionality only with slightly difference in accepted arguments, it will be much beneficial if we can unify them as soon as possible.

Description of the Change:

• devices/preprocess.py::_operator_decomposition_gen has been removed.
• Instead, a unify _operator_decomposition_gen living in transforms/decompose.py to have the same core internal implementation for _operator_decomposition_gen.
• In comparison, the new _operator_decomposition_gen accepts two new args:
  • custom_decomposer : to support the decomposer feature of the preprocess.decompose, which was originally supported by the removed one.
  • strict : If True, an error will be raised when an operator does not provide a decomposition and does not meet the stopping criteria. Otherwise, a warning is raised.

Benefits:
Avoid code duplication, promotes feature parity.

Possible Drawbacks:
The _operator_decomposition_gen is getting bulky.

Related GitHub Issues:
[sc-98684]

[github-actions]

Hello. You may have forgotten to update the changelog!
Please edit doc/releases/changelog-dev.md with:

• A one-to-two sentence description of the change. You may include a small working example for new features.
• A link back to this PR.
• Your name (or GitHub username) in the contributors section.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.69%. Comparing base (93867ba) to head (7e057e0).
⚠️ Report is 3 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #8193      +/-   ##
==========================================
- Coverage   99.69%   99.69%   -0.01%     
==========================================
  Files         548      548              
  Lines       56851    56845       -6     
==========================================
- Hits        56678    56672       -6     
  Misses        173      173              

☔ 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.

[JerryChen97]

@astralcai Hopefully I didn't resolve the merge conflicts with wrongly changing anything you wrote before

[JerryChen97]

I would say actually this PR looks already ready for review. Would you like to suggest what is remained to implement here? @astralcai

[astralcai]

I would say actually this PR looks already ready for review. Would you like to suggest what is remained to implement here? @astralcai

I don't think there's anything left to do here, but since nobody has capacity to take on more reviews this week, it'll likely have to wait till next week. I'll leave it up to you to address any review comments you receive.

[andrijapau]

That was pretty straight forward. Thanks! LGTM.