> insns.def:857:5: error: assigning to 'rb_zjit_func_t' (aka 'unsigned
> long (*)(struct rb_execution_context_struct *, struct
> rb_control_frame_struct *, unsigned long (*)(struct
> rb_execution_context_struct *, struct rb_control_frame_struct *))') from
> 'void *' converts between void pointer and function pointer
> [-Werror,-Wpedantic]

Thanks to Kokubun for noticing.

Follow-up: b0711b1cf152afad0a480ee2f9bedd142a0d24ac

Suppose YJIT runs a rb_vm_opt_send_without_block()
fallback and the control frame stack looks like:

```
will_tailcall_bar [FINISH]
caller_that_used_fallback
```

will_tailcall_bar() runs in the interpreter and sets up a tailcall.
Right before JIT_EXEC() in the `send` instruction, the stack will look like:

```
bar [FINISH]
caller_that_used_fallback
```

Previously, JIT_EXEC() ran bar() in JIT code, which caused the `FINISH`
flag to return to the interpreter instead of to the JIT code running
caller_that_used_fallback(), causing code to run twice and probably
crash. Recent flaky failures on CI about "each stub expects a particular
iseq" are probably due to leaving methods twice in
`test_optimizations.rb`.

Only run JIT code from the interpreter if a new frame is pushed.

I don't use that these days. I'd like to remove this to ease a change
I'm writing. I'll resurrect this when I need this again.

I closed https://github.com/ruby/ruby/pull/7543, but part of the diff
seems useful regardless, so I extracted it.