Add F64Const32 instruction

[Robbepop]

This is an optimization for certain constant f64 values that can losslessly be encoded as 32-bit f32 values. This way we can avoid allocating ConstRef for many common cases.

[paritytech-cicd-pr]

BENCHMARKS

	NATIVE			WASMTIME
BENCHMARK	MASTER	PR	DIFF	MASTER	PR	DIFF	WASMTIME OVERHEAD
execute/ bare_call_0	1.52ms	1.52ms	⚪ 0.31%	1.14ms	1.28ms	🔴 16.03%	🟢 -16%
execute/ bare_call_0/typed	1.13ms	1.12ms	⚪ -1.41%	712.17µs	790.42µs	🔴 8.94%	🟢 -29%
execute/ bare_call_1	1.59ms	1.59ms	⚪ 0.28%	1.34ms	1.45ms	🔴 8.84%	🟢 -9%
execute/ bare_call_16	2.57ms	2.51ms	⚪ -1.25%	3.70ms	3.70ms	🔴 0.85%	🟢 48%
execute/ bare_call_16/typed	1.54ms	1.58ms	⚪ 1.44%	1.92ms	2.09ms	🔴 10.48%	🟢 33%
execute/ bare_call_1/typed	1.24ms	1.24ms	⚪ -0.49%	1.06ms	1.12ms	🔴 4.86%	🟢 -9%
execute/ bare_call_4	1.76ms	1.76ms	⚪ -0.58%	1.84ms	2.02ms	🔴 8.79%	🟢 15%
execute/ bare_call_4/typed	1.23ms	1.23ms	⚪ 0.11%	1.02ms	1.11ms	🔴 8.02%	🟢 -10%
execute/ br_table	1.37ms	1.33ms	🟢 -2.51%	1.25ms	1.25ms	⚪ -0.99%	🟢 -6%
execute/ count_until	620.65µs	645.17µs	🔴 4.09%	1.42ms	1.45ms	⚪ 1.91%	🔴 124%
execute/ factorial_iterative	318.64µs	319.53µs	⚪ 0.21%	727.43µs	723.61µs	⚪ 1.57%	🔴 126%
execute/ factorial_recursive	509.68µs	514.67µs	⚪ 0.98%	1.06ms	934.86µs	🟢 -11.10%	🟡 82%
execute/ fibonacci_iter	1.40ms	1.40ms	⚪ -0.07%	3.41ms	3.50ms	⚪ 2.48%	🔴 151%
execute/ fibonacci_rec	4.21ms	4.22ms	⚪ 0.16%	8.20ms	8.39ms	⚪ -1.04%	🟡 99%
execute/ fibonacci_tail	972.38µs	1.08ms	🔴 11.28%	1.96ms	1.95ms	⚪ -0.62%	🟡 81%
execute/ global_bump	741.87µs	745.77µs	⚪ 0.62%	1.91ms	1.83ms	🟢 -5.00%	🔴 145%
execute/ global_const	660.69µs	712.87µs	🔴 7.82%	1.88ms	1.93ms	🔴 3.36%	🔴 171%
execute/ host_calls	36.45µs	37.54µs	⚪ 2.03%	42.27µs	48.43µs	🔴 13.74%	🟢 29%
execute/ memory_fill	1.15ms	1.16ms	⚪ 0.22%	2.89ms	2.84ms	🟢 -2.18%	🔴 145%
execute/ memory_sum	1.13ms	1.13ms	⚪ 0.00%	2.69ms	2.80ms	🔴 3.57%	🔴 147%
execute/ memory_vec_add	2.37ms	2.30ms	⚪ -2.85%	5.86ms	6.14ms	🔴 5.55%	🔴 167%
execute/ recursive_is_even	713.00µs	708.66µs	⚪ -0.58%	1.49ms	1.39ms	🟢 -7.71%	🟡 96%
execute/ recursive_ok	103.61µs	102.97µs	⚪ 0.02%	204.18µs	199.67µs	🟢 -2.60%	🟡 94%
execute/ recursive_scan	140.49µs	136.65µs	🟢 -2.80%	253.22µs	261.59µs	🔴 2.99%	🟡 91%
execute/ recursive_trap	10.14µs	10.13µs	⚪ -0.11%	20.96µs	20.31µs	⚪ -1.54%	🔴 101%
execute/ regex_redux	466.98µs	467.49µs	⚪ -0.54%	967.31µs	1.04ms	🔴 8.76%	🔴 123%
execute/ rev_complement	426.87µs	433.63µs	⚪ 0.71%	994.09µs	1.04ms	🔴 3.44%	🔴 139%
execute/ tiny_keccak	316.51µs	345.53µs	🔴 8.51%	889.58µs	883.38µs	⚪ 1.27%	🔴 156%
execute/ trunc_f2i	741.48µs	739.47µs	⚪ -0.67%	1.90ms	2.03ms	🔴 6.33%	🔴 175%
instantiate/ wasm_kernel	53.16µs	53.09µs	⚪ 0.60%	60.72µs	68.26µs	🔴 8.07%	🟢 29%
translate/ erc1155	187.37µs	186.29µs	⚪ -0.81%	337.59µs	325.46µs	🟢 -3.37%	🟡 75%
translate/ erc20	91.57µs	91.38µs	⚪ 0.13%	171.13µs	163.67µs	🟢 -5.05%	🟡 79%
translate/ erc721	130.14µs	128.01µs	🟢 -1.71%	238.34µs	229.36µs	⚪ -2.71%	🟡 79%
translate/ spidermonkey	56.33ms	55.98ms	⚪ -0.69%	0.00ns	0.00ns	⚪ -1.08%	🟢 -100%
translate/ wasm_kernel	3.75ms	3.72ms	⚪ -0.65%	6.39ms	6.18ms	🟢 -3.97%	🟡 66%

Link to pipeline