Eigenvalue: add the ArnoldiMethodJL keyword constructor

[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

• ArnoldiMethodJL is exported but had no keyword constructor, only the positional ArnoldiMethodJL(::NamedTuple). Its siblings all have one, so this adds the matching ArnoldiMethodJL(; kwargs...).
• Using this backend requires using ArnoldiMethod, which binds ArnoldiMethod to the module, so a bare ArnoldiMethod(; kwargs...) then fails with objects of type Module are not callable. With no keyword constructor on the JL name either, a using LinearSolve user had no working unqualified spelling.
• Nothing renamed or removed. LinearSolve.ArnoldiMethod(...) still works and the two produce identical objects, which the new test asserts.
• Every existing assertion in test/Core/eigenvalue.jl uses the qualified form, which is why this was never caught. The new test uses the unqualified one.
• Verified on a 50x50 SPD matrix: the new constructor forwards kwargs and its eigenvalues match eigvals to 8 significant digits. test/Core/eigenvalue.jl passes, Runic clean.
• Found while checking Direct and Iterative eigensolvers #487, but this does not close it: the eigensolvers are already implemented, and the ModelingToolkit preconditioner interface that issue also asks for is not built.

AI Disclosure: Used Opus 5

[AJ0070]

@ChrisRackauckas