Simplify floating points comparison for Object.is

[MaxGraey]

This change significantly simplify Object.is<f64 | f32>.

Sample 1:

Object.is(x, -0.0);

OLD codegen:

  local.get $x
  i64.reinterpret_f64
  i64.const -9223372036854775808
  i64.eq
  local.get $x
  f64.const -0
  f64.eq
  i32.and
 )

NEW codegen:

 local.get $x
 i64.reinterpret_f64
 i64.const -9223372036854775808
 i64.eq

Sample 2:

Object.is(x, NaN);

OLD codegen:

  local.get $x
  i64.reinterpret_f64
  i64.const 9221120237041090560
  i64.eq
  local.get $x
  local.get $x
  f64.ne
  local.get $x
  f64.const nan:0x8000000000000
  f64.eq
  select

NEW codegen:

 local.get $x
 i64.reinterpret_f64
 i64.const 9221120237041090560
 i64.eq
 local.get $x
 local.get $x
 f64.ne
 i32.or

• I've read the contributing guidelines
• I've added my name and email to the NOTICE file