stubtest: don't crash when a classmethod's first parameter isn't named cls

[HuzaifaChaudary]

fixes #16583

the first parameter of a classmethod can be called anything . networkx really does write
def construct(EdgeComponentAuxGraph, G) . stubtest checked that name against a list of four
and raised StubtestFailure when it did not match :

StubtestFailure: unexpected class parameter name 'EdgeComponentAuxGraph' in pkg.EdgeComponentAuxGraph.construct

that is not a normal stubtest error , it stops the whole run , so one oddly named parameter
takes out stub checking for a whole distribution . that is what happens on
python/typeshed#13587 .

the name is never used for anything . the line right under it drops the argument by position ,
so the check was not guarding the code below it . it looks at the argument kind now instead .

while testing i found a second way onto the same line . a stub written def f(*args) under
@classmethod also crashed . there is nothing bound to drop in that case , and
inspect.signature of the runtime classmethod keeps the star argument too , so both sides
already match and it returns unchanged . dropping it would have made a false mismatch .

two cases added to test_static_class_method , one for the odd name and one for *args .
both fail on master with the crash above .

teststubtest.py and teststubgen.py give 440 passed , 1 skipped , 2 xfailed . self check
on the two files is clean , and so are black and ruff .

small heads up , #21863 adds its own cases to test_static_class_method starting on the same
line , so whichever goes in second will need a rebase . it is only the test file , that one
changes verify_var and this one changes _resolve_funcitem_from_decorator , so they do not
touch the same code .

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.