Merge raw_bitcast and bitcast (#5175)

- Allow bitcast for vectors with differing lane widths
- Remove raw_bitcast IR instruction
- Change all users of raw_bitcast to bitcast
- Implement support for no-op bitcast cases across backends

This implements the second step of the plan outlined here:
https://github.com/bytecodealliance/wasmtime/issues/4566#issuecomment-1234819394

cranelift/filetests/filetests/runtests/simd-bitcast.clif

@@ -1,23 +1,8 @@
 test interpret
 test run
 target aarch64
-; x86_64 and s390x do not support vector bitcasts.
-
-function %bitcast_if32x2(i32x2) -> f32x2 {
-block0(v0: i32x2):
-  v1 = bitcast.f32x2 v0
-  return v1
-}
-; run: %bitcast_if32x2([0 4294967295]) == [0x0.0 -NaN:0x3fffff]
-; run: %bitcast_if32x2([-1 127]) == [-NaN:0x3fffff 0x0.0000fep-126]
-
-function %bitcast_fi32x2(f32x2) -> i32x2 {
-block0(v0: f32x2):
-  v1 = bitcast.i32x2 v0
-  return v1
-}
-; run: %bitcast_fi32x2([0x0.0 -NaN:0x3fffff]) == [0 4294967295]
-; run: %bitcast_fi32x2([-NaN:0x3fffff 0x0.0000fep-126]) == [-1 127]
+target x86_64
+target s390x
 
 function %bitcast_if32x4(i32x4) -> f32x4 {
 block0(v0: i32x4):