Refactor containers h includes

[RikuMinamiyama]

Overview
Removed unnecessary CPP and C header includes.

Correction details

• C header removal

  The following modules use values defined in containers.h:

  /haskell-containers/containers/src/Data/Graph.hs
  → DEFINE_PATTERN_SYNONYMS, USE_ST_MONAD, USE_UNBOXED_ARRAYS
  
  /haskell-containers/containers/src/Data/Sequence.hs
  → DEFINE_PATTERN_SYNONYMS
  
  /haskell-containers/containers/src/Data/IntSet/Internal.hs
  → WORD_SIZE_IN_BITS
  
  /haskell-containers/containers/src/Data/Map/Internal.hs
  → WORD_SIZE_IN_BITS
  
  /haskell-containers/containers/src/Data/Sequence/Internal.hs
  → DEFINE_PATTERN_SYNONYMS
  
  /haskell-containers/containers-tests/tests/IntSetValidity.hs
  → WORD_SIZE_IN_BITS
  

  Removed #include "containers.h" from all other modules.

• CPP removal

  Removed CPP language pragmas from files that do not use CPP directives.

Testing

• cabal build containers succeeds
• cabal test containers-tests passes all tests
• cabal bench containers-tests passes all benchmarks

While the changes are straightforward, they affect a wide range of files, so careful review is requested.

Fixes #968

[meooow25]

Thanks for the PR, I'm taking a look but I would like to know, were LLMs were involved in this?

[RikuMinamiyama]

Thank you for your confirmation.

Regarding containers.h , I searched for values defined in containers.h , such as DEFINE_PATTERN_SYNONYMS , and removed the containers.h include from files where these values were absent. (LLMs were not used)

Regarding C++ pragmas, I first checked each file for the presence of #ifdef and similar directives. (No LLM was used.) Afterward, out of concern, I asked the LLM if there were any places where C++ pragma declarations might have been accidentally left in.

Additionally, when drafting the PR description, I used an LLM to translate from Japanese to English.

If there are any errors or omissions, please provide guidance.

[meooow25]

Thanks, I don't mind if you use an LLM for translation.
But I would prefer to not review LLM-generated PRs, particularly for low-hanging-fruit where it would miss the whole point of introducing the code to a human contributor.

[meooow25]

I think map-strictness.hs could also do without the {-# LANGUAGE CPP #-}, would you please add that to the PR?
Also, it needs a rebase to avoid conflicts.