When we call `asm.load`, many times we're passing in a VReg, and that
causes extra loads when we lower to machine code.  I'd like to only emit
a load in the case that the operand _isn't_ a VReg.

For example this code:

```ruby
class Foo
  def initialize
    @foo = 123
  end

  def foo
    @foo
  end
end

foo = Foo.new
5.times { foo.foo }
```

Before this patch, the machine code for `LoadField` looks like this:

```
  # Insn: v18 LoadField v17, :_shape_id@0x4
  # Load field id=_shape_id offset=4
  0x121308320: mov x1, x0
  0x121308324: ldur w1, [x1, #4]
```

Now it looks like this:

```
  # Insn: v18 LoadField v17, :_shape_id@0x4
  # Load field id=_shape_id offset=4
  0x12339c320: ldur w1, [x0, #4]
```

We were able to eliminate a reg-reg copy.

pm_parse_process initializes the index_lookup_table but nothing seems to
use it after it has been allocated. However, pm_compile_scope_node will
overwrite the index_lookup_table and cause it to leak memory. This can
be seen during bootup with the following memory leaks reported by ASAN:

    #0 0x60dba31b7af3 in malloc
    #1 0x60dba32e0718 in rb_gc_impl_malloc gc/default/default.c:8287:5
    #2 0x60dba32c7aa7 in ruby_xmalloc_body gc.c:5373:12
    #3 0x60dba32c4a54 in ruby_xmalloc gc.c:5355:34
    #4 0x60dba3260314 in pm_index_lookup_table_init_heap prism_compile.h:89:29
    #5 0x60dba3209388 in pm_parse_process prism_compile.c:11366:5