Add LinearSolvePyAMG sub-library with PyAMG integration and tests

[AJ0070]

Checklist

• Appropriate tests were added
• Any code changes were done in a way that does not break public API
• All documentation related to code changes were updated
• The new code follows the
  contributor guidelines, in particular the SciML Style Guide and
  COLPRAC.
• Any new documentation only uses public API

Additional context

Changes

• lib/LinearSolvePyAMG/ : new sub-library (Project.toml, CondaPkg.toml, src/, test/)
• docs/src/basics/Preconditioners.md : replaced stale cortner/PyAMG.jl entry
• docs/src/solvers/solvers.md : added ### LinearSolvePyAMG.jl section
• test/basictests.jl : added opt-in @static if isdefined(...) test block
• .gitignore : added .CondaPkg/

Implemented as a sub-library (like lib/LinearSolveAutotune/) rather than a weak dependency because PyAMG is not in the Julia General registry.

Resolves #372

AI disclosure: used claude and gemini to help in replicating structure of LinearSolveAutotune in LinearSolvePyAMG, adding comments and refactoring documentation.

[AJ0070]

I created a separate test group for PyAMG to prevent the extra setup time from installing Python and PyAMG from slowing down every Core CI job. If there is a better way to handle this, I would appreciate your thoughts @ChrisRackauckas.

[ChrisRackauckas]

No AI was used at all?

[AJ0070]

Sorry, I missed it. I have added it now and will keep it in mind to include it.

[AJ0070]

Are there any blockers with this?

[AJ0070]

@ChrisRackauckas

[ChrisRackauckas]

Just the naming fix

[AJ0070]

fixed it

[AJ0070]

@ChrisRackauckas are there any other high priority issues you would suggest I focus on?

[ChrisRackauckas]

In the same vein, #170, #47, #182, and #51

#70 would then be a good direction to go for a bit.

Then #143 #487 #561 tackling EigenProblem would be a longer term but major direction to go.