Add prototype inline cache

[ErosZy]

I add prototype inline cache for basic get/set op, there are benchmark results:

• Apple M1 Pro
• macOS 14.4.1
• Clang 15.0.0 arm64-apple-darwin23.4.0

	prototype-inline-cache	master	Improvement(%)
Richards	1363	1263	+7.91
DeltaBlue	1346	1197	+12.44
Crypto	1258	1227	+2.52
RayTrace	1370	1307	+4.82
EarleyBoyer	2340	2344	-0.29
RegExp	277	270	+2.59
Splay	2537	2307	+9.96
SplayLatency	9276	9214	+0.67
NavierStokes	2251	2197	+2.45
PdfJS	4345	4265	+1.88
Mandreel	1055	1031	+2.32
MandreelLatency	7163	7081	+1.15
Gameboy	9057	9034	+0.25
CodeLoad	17284	16973	+0.19
Box2D	5433	5234	+3.80
Typescript	17650	17766	-0.65
Score (version 9)	3041	2944	+3.29

• Intel i5-12400F
• Windows 10 - mingw
• GCC 13.2.0

	prototype-inline-cache	master	Improvement(%)
Richards	845	686	+23.17
DeltaBlue	679	575	+18,08
Crypto	801	841	-2.52
RayTrace	450	422	+4.75
EarleyBoyer	871	884	-1.47
RegExp	161	157	+2.54
Splay	950	945	+0.52
SplayLatency	4743	4564	+3.92
NavierStokes	1243	1240	+0.24
PdfJS	1942	1839	+5.60
Mandreel	532	546	-2.56
MandreelLatency	3655	3553	+1.15
Gameboy	4671	4336	+2.87
CodeLoad	8613	8485	+1.50
Box2D	2552	2171	+17.54
Typescript	8892	8379	+6.12
Score (version 9)	1485	1411	+5.24

• Microsoft AMD-V(CPU MHz: 3244.657)
• Linux 6.8.0-1014-azure
• gcc (Ubuntu 9.4.0-1ubuntu1~20.04.2) 9.4.0

	prototype-inline-cache	master	Improvement(%)
Richards	750	657	+14.15
DeltaBlue	746	604	+23.50
Crypto	614	603	+1.82
RayTrace	834	840	-0.71
EarleyBoyer	1216	1256	-3.18
RegExp	202	202	+0.00
Splay	1680	1618	+3.83
SplayLatency	5634	4945	+13.93
NavierStokes	1022	1004	+1.79
PdfJS	2369	2246	+5.47
Mandreel	478	451	+5.98
MandreelLatency	3408	3110	+9.58
Gameboy	4303	3928	+9.54
CodeLoad	11991	11628	+3.12
Box2D	2535	2332	+8.70
Typescript	7862	7934	-0.90
Score (version 9)	1639	1548	+6.84

[saghul]

Nice!

[ErosZy]

@saghul follow #244, I've also added the benchmark test results for Linux and Windows MinGW here for reference.

[saghul]

Thank you! Not sure I am qualified 😅 but I'll give it a try!

[saghul]

@bnoordhuis Could you PTAL? LGTM in so far as I can read it and the tests pass 😅

[bnoordhuis]

I'll try to take a look over the weekend.

[chqrlie]

@bnoordhuis Could you PTAL? LGTM in so far as I can read it and the tests pass 😅

I shall take a look over the week-end too. I have an alternate version on the back burner so let's see if @ErosZy approach is promissing...

[ErosZy]

@bnoordhuis @saghul I've made some changes to the code based on the review; could you please take another look? Thanks.

[bnoordhuis]

Mostly looks good but I still have a few comments.

[bnoordhuis]

@ErosZy heads up, I'm doing some IC refactoring in #609 because I'm fairly sure I've identified some bugs. You may have to rework your changes a little.

(I wouldn't normally push such changes when I know they conflict with pull requests from contributors but bug fixes trump enhancements.)

[p-bakker]

Any chance in this one being finished and included in the forthcoming release? Any performance bump is always great to have 🙂

[saghul]

Converting to draft while the feedback is addressed. It also needs a rebase...

[ErosZy]

@saghul @bnoordhuis please check, thx.

[saghul]

Tests are 💚 ! @bnoordhuis can you PTAL?

[ErosZy]

add a screenshot of the benchmark, and after merging with the latest master branch, it still maintains a good performance improvement. It did not cause performance degradation. 😊