Refactor Engine executor to handle Wasm calls and returns

[Robbepop]

No description provided.

[paritytech-cicd-pr]

BENCHMARKS

	NATIVE			WASMTIME
BENCHMARK	MASTER	PR	DIFF	MASTER	PR	DIFF	WASMTIME OVERHEAD
execute/ bare_call_0	1.44ms	1.87ms	🔴 30.44%	1.07ms	1.34ms	🔴 25.82%	🟢 -28%
execute/ bare_call_0/typed	1.02ms	1.12ms	🔴 9.51%	737.21µs	945.52µs	🔴 28.29%	🟢 -15%
execute/ bare_call_1	1.49ms	1.84ms	🔴 22.63%	1.18ms	1.40ms	🔴 18.09%	🟢 -24%
execute/ bare_call_16	2.32ms	2.49ms	🔴 6.17%	3.12ms	3.30ms	🔴 5.84%	🟢 33%
execute/ bare_call_16/typed	1.59ms	1.79ms	🔴 11.73%	1.52ms	1.55ms	🔴 2.14%	🟢 -13%
execute/ bare_call_1/typed	1.14ms	1.22ms	🔴 7.46%	986.90µs	1.22ms	🔴 23.25%	🟢 -1%
execute/ bare_call_4	1.56ms	1.75ms	🔴 12.20%	1.53ms	1.88ms	🔴 22.88%	🟢 7%
execute/ bare_call_4/typed	1.12ms	1.26ms	🔴 12.40%	1.03ms	1.14ms	🔴 10.94%	🟢 -9%
execute/ br_table	1.16ms	1.30ms	🔴 11.56%	1.21ms	1.38ms	🔴 14.68%	🟢 6%
execute/ count_until	650.11µs	680.20µs	🔴 4.65%	1.21ms	1.13ms	🟢 -7.01%	🟡 66%
execute/ factorial_iterative	321.56µs	335.81µs	🔴 4.46%	505.40µs	477.58µs	🟢 -5.47%	🟢 42%
execute/ factorial_recursive	622.92µs	595.73µs	🟢 -4.52%	1.09ms	953.44µs	🟢 -12.96%	🟡 60%
execute/ fib_iterative	1.46ms	1.47ms	⚪ 0.41%	2.45ms	2.23ms	🟢 -9.17%	🟡 52%
execute/ fib_recursive	5.74ms	4.80ms	🟢 -16.33%	9.70ms	7.76ms	🟢 -19.90%	🟡 62%
execute/ global_bump	892.73µs	823.45µs	🟢 -7.89%	2.26ms	1.56ms	🟢 -31.07%	🟡 90%
execute/ global_const	685.20µs	739.16µs	🔴 7.78%	1.45ms	1.43ms	🟢 -1.49%	🟡 94%
execute/ host_calls	27.48µs	31.72µs	🔴 15.31%	38.98µs	49.40µs	🔴 26.74%	🟡 56%
execute/ memory_fill	1.20ms	1.20ms	⚪ 0.32%	2.09ms	1.93ms	🟢 -7.40%	🟡 61%
execute/ memory_sum	1.26ms	1.18ms	🟢 -7.32%	2.14ms	1.93ms	🟢 -9.58%	🟡 63%
execute/ memory_vec_add	2.39ms	2.45ms	🔴 2.55%	4.91ms	4.59ms	🟢 -6.53%	🟡 87%
execute/ recursive_is_even	1.15ms	1.03ms	🟢 -10.10%	1.90ms	1.69ms	🟢 -10.73%	🟡 64%
execute/ recursive_ok	145.13µs	122.65µs	🟢 -15.38%	261.08µs	203.63µs	🟢 -22.02%	🟡 66%
execute/ recursive_scan	176.56µs	159.13µs	🟢 -9.88%	311.88µs	250.44µs	🟢 -19.74%	🟡 57%
execute/ recursive_trap	14.25µs	12.54µs	🟢 -11.94%	26.21µs	21.96µs	🟢 -16.12%	🟡 75%
execute/ regex_redux	530.13µs	520.04µs	🟢 -2.13%	1.01ms	911.31µs	🟢 -9.80%	🟡 75%
execute/ rev_complement	463.03µs	460.88µs	⚪ -0.46%	838.65µs	782.77µs	🟢 -6.64%	🟡 70%
execute/ tiny_keccak	328.92µs	324.64µs	🟢 -1.23%	641.50µs	603.33µs	🟢 -5.98%	🟡 86%
execute/ trunc_f2i	765.31µs	782.74µs	🔴 2.38%	1.44ms	1.40ms	🟢 -2.48%	🟡 79%
instantiate/ wasm_kernel	54.14µs	54.64µs	⚪ 2.04%	56.55µs	59.58µs	🔴 6.63%	🟢 9%
translate/ erc1155	234.92µs	233.67µs	⚪ -0.56%	441.19µs	438.96µs	⚪ -0.67%	🟡 88%
translate/ erc20	113.57µs	113.75µs	⚪ 0.30%	213.79µs	211.15µs	⚪ -1.08%	🟡 86%
translate/ erc721	160.91µs	161.78µs	⚪ 0.56%	311.76µs	307.21µs	🟢 -1.61%	🟡 90%
translate/ spidermonkey	0.00ns	0.00ns	⚪ 1.33%	0.00ns	0.00ns	⚪ -0.21%	🟢 0%
translate/ wasm_kernel	4.14ms	4.17ms	⚪ 1.19%	8.23ms	8.16ms	⚪ -1.12%	🟡 96%

Link to pipeline

[codecov-commenter]

Codecov Report

Merging #694 (7906c9c) into master (43ce25d) will decrease coverage by 1.22%.
The diff coverage is 86.26%.

@@            Coverage Diff             @@
##           master     #694      +/-   ##
==========================================
- Coverage   81.01%   79.80%   -1.22%     
==========================================
  Files         100      100              
  Lines        8506     8203     -303     
==========================================
- Hits         6891     6546     -345     
- Misses       1615     1657      +42     

Impacted Files	Coverage Δ
crates/wasmi/src/engine/stack/values/sp.rs	95.18% <ø> (ø)
crates/wasmi/src/engine/code_map.rs	82.50% <50.00%> (-1.72%)	⬇️
crates/wasmi/src/engine/cache.rs	81.25% <71.42%> (-0.57%)	⬇️
crates/wasmi/src/engine/stack/mod.rs	74.64% <82.35%> (-7.18%)	⬇️
crates/wasmi/src/engine/mod.rs	81.72% <84.84%> (-2.14%)	⬇️
crates/wasmi/src/engine/stack/values/tests.rs	97.72% <87.50%> (-2.28%)	⬇️
crates/wasmi/src/engine/stack/values/mod.rs	86.84% <92.30%> (-0.34%)	⬇️
crates/wasmi/src/engine/executor.rs	86.52% <93.45%> (-8.56%)	⬇️
crates/wasmi/src/engine/stack/frames.rs	100.00% <100.00%> (ø)
... and 2 more

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[Robbepop]

With respect to benchmarks it is expected that host call performance regresses with this PR.
That said it is not (yet) clear to me why bare_calls, global_const, count_until and factorial_iterative regress slightly on native.
However, those minor regressions are outweight by the impressive performance gains on both native and Wasm especially for recursive work loads.
Therefore, I propose to finally merge this PR and research the aforementioned performance regressions in follow-up PRs.