Ruby

[ruby/rubygems] Address review feedback on tmp directory path resolution

Use source_root.parent instead of hardcoded relative path to derive the
repository root, so the code remains correct if the support file is ever
moved. Also fix the example Gemfile source path in DEBUGGING.md to...

[ruby/rubygems] Move bundler test tmp directory from bundler/tmp to top-level tmp

The bundler tests were writing temporary files to bundler/tmp, but this
moves them to the repository's top-level tmp directory instead, which is
already covered by .gitignore. All references in CI workflows, SimpleCov...

Bump the github-actions group across 1 directory with 3 updates

Bumps the github-actions group with 3 updates in the / directory: ruby/setup-ruby, lewagon/wait-on-check-action and taiki-e/install-action....

ZJIT: Profiling: Separate bits for T_CLASS and T_MODULE

So I don't need to use is_subclass_of(). We already have two branches
for the two types during profiling anyways.

ZJIT: Print the Cptr, not a pointer to the pointer

Previously, we had none-sensical lines such as:

v15:CPtr[CPtr(0x1008)] = Const CPtr(0x1010)

Which was a contradiction as the pointer in the type specialization
should match the pointer on the right hand side.

ZJIT: Print CUInt64 in hex when sign bit is set

It's likely some sort of mask or flags. Hexadecimal is more compact.

ZJIT: Guard T_* in addition to shape in polymorphic getivar

This is a8f3c34556 ("ZJIT: Add missing guard on ivar access on
T_{DATA,CLASS,MODULE}") but for the polymorphic implementation in HIR
build.

Add assertion to rb_class_alloc

We should only ever call rb_class_alloc with a T_CLASS. Adding this
assert/assumption allows us to elide later checks for T_CLASS>

Use RB_BUILTIN_TYPE in RCLASS_SINGLETON_P

This should avoid us emitting the check for a special constant.

Avoid box st_lookup in rb_class_real